The most common mistake I see in a machine setup is treating the job as “get the machine level.” It is not. When you put a precision level for machine setup on a lathe bed, you are correcting twist along the ways, not chasing absolute level. The goal is two matching readings at each end of the bed, not a centred bubble. Once you understand that, choosing the right machinist level becomes simple.
That matters because sensitivity alone is a poor guide to buying. A .0002 in/10 in level is the most sensitive class on the market and the least pleasant to use day to day. Shop talk across Practical Machinist, The Hobby-Machinist and Canadian Hobby Metal Workers keeps arriving at the same place: buy the level you will actually pick up, not the one with the best number on the box.
We pulled ten precision levels together for this roundup, from a 2-1/2 inch Starrett pocket level to two-axis digital machinist levels with a phone readout. Each one is rated in the machinist convention of inches per foot, inches per 10 inches, seconds of arc and millimetres per metre, so you can compare any spec sheet against any other. I have added the workflow, the length rules and the calibration method the forums actually use.
Table of Contents
Top 3 Picks for Machine Setup (October 2026)
Three of these ten earn a badge before we get into the details: the Starrett 98-6 as the all-round shop default, the Accusize 12 inch for full-bed work, and the Digi-Pas DWL1500XY if you want to level a machine alone.
Starrett 98-6 Machinists Level
- 0.005 in/ft sensitivity
- 0.42 mm/m
- 80 to 90 seconds of arc
- Adjustable ground vial in brass tube
Accusize S908-C687 12 inch Master
- 0.0002 in/10 in accuracy
- 0.02 mm/m
- V-groove base for round work
- Thermal barrier grip
Digi-Pas DWL1500XY Smart Level
- 0.0002 in/ft resolution
- 0.02 mm/m
- Two-axis Bluetooth readout
- NIST JIS DIN traceable certificate
Quick Overview: 10 Precision Levels in 2026
Every level below is written in the units machinists cross-check. Inches per foot and inches per 10 inches are not the same scale, and mixing them up is the single biggest source of confusion when reading a listing.
| Product | Specs | Action |
|---|---|---|
| Starrett 98-6 Machinists Level 6 inch |
|
Check Latest Price |
 Yan He 60mm Cardan Precision Level |
|
Check Latest Price |
 Starrett 135A Pocket Level 2-1/2 inch |
|
Check Latest Price |
| Accusize S908-C687 12 inch Master Level |
|
Check Latest Price |
| Digi-Pas DWL1500XY 2-Axis Smart Level |
|
Check Latest Price |
 findmall 8 Inch 0.0002 per 10 Inch Level |
|
Check Latest Price |
 Starrett 134 Cross Test Level and Plumb |
|
Check Latest Price |
 Accusize S908-C612 12 inch Master Level |
|
Check Latest Price |
 Digi-Pas DWL1300XY 2-Axis Smart Level |
|
Check Latest Price |
 WFLNHB 12 Inch 0.0002 per 10 Inch Level |
|
Check Latest Price |
What Accuracy Do You Actually Need?
Standard machinist levels run .005 in/ft, which is 0.42 mm/m or 80 to 90 seconds of arc. True master precision levels reach .0002 in/10 in, or 0.02 mm/m, roughly 4 to 5 seconds of arc. For regular machine setup, the practical ceiling is about .0005 in/10 in, because anything more sensitive is hard to keep settled long enough to read.
Here is the sensitivity ladder, with the four unit systems machinists use:
.005 in/ft equals 0.42 mm/m, roughly 80 to 90 seconds of arc. This is the Starrett 98 standard and it is plenty for leveling a lathe or mill base before grouting.
.002 in/ft equals about 0.18 mm/m, roughly 40 seconds of arc. A good all-round shop figure and a common digital readout resolution.
.0004 in/ft equals about 0.04 mm/m, roughly 10 seconds of arc. A typical stated accuracy for a fine digital machinist level.
.0002 in/10 in equals 0.02 mm/m, roughly 4 to 5 seconds of arc. The most sensitive common class, also called two tenths per foot.
Note the trap. .0002 in/10 in and .0002 in/ft are different sensitivities by a factor of twelve, yet both appear in listings with almost identical wording. Convert before you compare two spec sheets.
Read the ladder as a job ladder. The coarse end is for rough leveling and for sizing shim stacks before grout sets. The fine end is for the last pass on ways and for checking a surface plate.
There is a technique that turns the reading into a number you can act on. A .004 in/10 in reading does not just tell you which corner is high, it tells you how much shim to add. Multiply the reading difference by the span between your measuring points and you have a thickness in thousandths.
You Are Untwisting the Bed, Not Leveling the Machine
This is the consensus answer in the machining forums, and almost no product guide mentions it. A bed that is bent in one plane needs a precision straight edge, not a level. A bed that is twisted needs a level. Mixing up the two is how people end up shimming a machine into worse condition than they found it.
Twist means the bed slopes differently at the headstock end than at the tailstock end. The level does not need to read centre. It needs to read the same at both ends. Set it on the ways near the headstock, note the graduation, then move it to the tailstock end and note that reading. If they match, the bed is untwisted, whatever the bubble is doing in absolute terms.
That principle also takes the pressure off calibration anxiety. As long as you never reverse the level end for end during a single setup, an uncalibrated level still tells you whether two readings agree. Calibration matters when you move the tool around the shop, not when you compare two spots on one pass.
One more practical note from the forums: a two-collar test with a test bar can finish a dial-in without any level at all. A level makes the process much faster. It is a shortcut, not a licence.
The 10 Best Precision Levels for Machine Setup
1. Starrett 98-6 Machinists Level, 6 Inch, Ground Vial and Cross Test
Starrett Machinists’ Level, Ground Main and Cross Test Vial, 6″ Length
0.005 in/ft sensitivity
0.42 mm/m, 80-90 seconds of arc
6 in L x 1.19 in W, cast iron
0.6 kg, made in USA
Pros
- Precision-machined cast iron base
- Adjustable vial with knurled knob
- Cross test vial levels two axes at once
- Easy to calibrate in 5 to 10 minutes
- Holds resale value
Cons
- Must be calibrated by the user before first use
- Some sizes ship without a protective case
- Occasional reports of out-of-calibration arrival
- Listing description reads as confusing
This is the level most shops end up with, and the one I recommend to anyone setting up a lathe or a mill for the first time. The 98-6 pairs a ground main vial with a cross test vial mounted at 90 degrees, so you can read both axes without rotating the tool. The vial sits in a knurled brass tube with a friction-fit cover, and the main vial is adjustable through a side knob.
Owners consistently describe the accuracy as a clear step above ordinary construction levels, with readings well beyond what a normal spirit level resolves. Typical uses reported are leveling machine platen beds and ways, and the tool is described as something you buy once and keep for decades.
The base is cast iron, precision-machined on the reference surface, and there is an involute groove between the bearing flats so the level centres itself on round work and shafts. That groove matters more than most listings admit, because a large share of placement errors on a lathe come from setting the tool down slightly canted.
Two things worth knowing before you buy. The vial is user-adjustable, so a first-use calibration is part of normal ownership, and reviewers put that at five to ten minutes on a verified flat surface. Some sizes are sold without a case, and a few owners report units arriving outside calibration, so read the vial on a surface plate before you trust a number.
How this reads at .005 in/ft in real terms
At 80 to 90 seconds of arc and 0.42 mm/m, one graduation shift equals about .005 in/ft of slope. Across a 40 inch bed that is roughly .017 in of difference end to end, which is comfortably inside what shim stock and grinding can absorb.
That makes it a rough-to-medium instrument rather than a finishing tool. It gets a machine close and tells you which corner is high. For the last pass on a surface plate you would want something finer.
Whether it suits your setup
This is the right pick if you are leveling a bed before grouting, aligning a headstock and tailstock, or re-checking a base after a move. It is the wrong pick if your only goal is a sub-.001 in readout across a wide span.
The same family comes in 8 inch and 12 inch bodies for wider beds. The 6 inch model is the easiest to move around and the easiest to keep clear of swarf, which is why it is the most commonly bought length.
2. Yan He 60mm Cardan Precision Level, 15 ft per 2 mm
High Precision Cardan Level with Horizontal Bubble – 60MM Accuracy, 15’/2
Accuracy 2 mm per 15 ft, about 0.44 mm/m
2.36 in L x 0.5 in W, 80 g
Aluminium housing with plastic vial
Symmetric bubble deflection
Pros
- Reads close to machinist level accuracy in a pocket format
- Machined aluminium housing feels solid
- Large clear lens with crisp dark graduations
- Compact enough to keep in a tool bag
Cons
- Higher gradations understate the true angle slightly
- Not for use in strong direct sunlight
- No accessories included
At 60 mm long and 80 g, this is not a bed level. It is the small check that stays in your pocket while truing work, setting a fixture or checking a machine component, where a full machinist level is too much tool. The stated accuracy is 2 mm per 15 ft, which converts to roughly 0.017 in/ft, about 0.44 mm/m, or near 90 seconds of arc.
That is a meaningful figure for a tool this size. It lands close to the Starrett 98 sensitivity band while fitting in a drawer, and owners compare it favourably to levels they already trust as go-to references.
The cardan hinge lets it sit on an angled surface without a wedge, which is the whole reason this design exists. The bubble is high contrast with a wide temperature range, and the large clear top lens with dark graduations is what makes it readable in poor shop lighting.
Two caveats from owners. Higher graduations tend to understate the true angle, so treat the top of the scale as approximate. And it is not meant for direct sun, which matters on an outdoor install and rarely matters indoors.
Where this one earns its place in a setup
It belongs beside the machine, not on it. Use it to check a subplate, a riser, a fixture face or a casting face before you bring out the machinist level and start the real work.
Buyers use it for turntables, gimbals, telescopes and general machine and appliance work. At this weight and size, keeping one in the tool bag costs nothing and saves borrowing the good level.
Whether to skip it for machine setup
If your task is leveling a lathe bed or a mill base, this is too short to bridge the points that matter. A 60 mm span cannot tell you anything about twist over 40 inches.
Buying it alongside a machinist level makes sense. Buying it instead of one does not, however good the accuracy looks on the listing.
3. Starrett 135A Pocket Level, 2-1/2 Inch
Starrett Pocket Level, 2-1/2″ Length
2-1/2 in length
Main vial included
Nickel-plated hex stock with convex ends
0.03 kg
Pros
- Pocketable so it is always on hand
- Starrett accuracy in a 2-1/2 in body
- Convex hex ends seat on round and machined surfaces
- Satin plating resists wear and corrosion
Cons
- Short span limits usefulness on long reference planes
- Not a substitute for a full length level on a machine bed
The 135A is the level that lives in a shirt pocket. It is machined from hexagonal stock with convex ends and a satin nickel-plated finish, and it carries a main vial as standard. At 2-1/2 inches and just over an ounce, it disappears into a tool roll.
Owners describe it as a long-standing shop favourite for quick checks on machine parts, small assemblies and round surfaces, with accuracy beyond a construction level within its sensitivity range. The convex ends are the detail worth knowing, because they let it sit on a round shaft without rocking.
The finish is satin nickel plating rather than bare cast iron, which is why it does not need the petroleum jelly storage treatment that cheap carbon steel imports demand. That makes it a sensible first level for a shop with no climate-controlled cabinet.
It is a genuinely accurate instrument, just a short one. The only recurring complaint in reviews is size, and owners are clear that it is not meant for long machine beds.
When a short level is the correct tool
Span the problem the other way. When the two points you are comparing are close together, a short level is more accurate in practice because there is less base to rock and less surface to read.
That covers fixture and vice alignment, checking a mounting face, and confirming a shaft or a gear sits square. On those jobs the 2-1/2 inch body is an advantage rather than a compromise.
How it fits into a larger setup kit
Think of it as the tool between the coarse pass and the fine pass. Get close with a frame or torpedo level, check with this, finish with a machinist level or a two-collar test.
Plenty of shops buy one of these after the machinist level, once it is clear how many questions are about a short span rather than a long bed.
4. Accusize S908-C687 12 Inch Master Precision Level, 0.0002 in/10 in
Accusize Industrial Tools Master Precision Level Accuracy, 12in, S908-C687
0.0002 in/10 in accuracy
0.02 mm/m adjustable main vial
12 in L x 1.81 in W alloy steel
2.75 kg, ANSI and DIN 887
Pros
- V-groove base seats on shafts without rocking
- Thermal barrier grip limits hand heat drift
- Ground base facing for granite and bed contact
- Built to ANSI and DIN 887 master level specs
Cons
- Heavier than comparable machinist levels
- Some units need calibration out of the box
This is the long-bed option. At 12 inches it spans most of a standard lathe bed in one placement, which means fewer repositionings and fewer chances to introduce a reading error. The advertised sensitivity is .0002 in/10 in, equal to 0.02 mm/m, about 4 to 5 seconds of arc.
Owners using it for lathe and mill alignment report that the advertised sensitivity holds up in practice. The V-groove base is the feature they mention most, because it seats on cylindrical pipes and rotating shafts without the rocking you get on a flat-only base.
The construction detail that separates it from cheaper copies is the thermal barrier handle. It isolates the frame from body heat, which limits thermal expansion during a long hand-held session. On a level this sensitive, warmth from your hand is a real source of drift.
The base is precision ground for flat contact on machine beds and granite reference tables, and the frame is specified against ANSI and DIN 887 master level standards. The trade-offs are honest ones: at 2.75 kg it is heavy for a hand tool, and some units need calibration before use.
When you need a 12 inch span instead of a 6 inch
Span is the deciding factor. If the two points you need to compare are further apart than your level is long, you are measuring in two overlapping segments and introducing error at every overlap. Measure first, then buy.
A 12 inch body also sits further from the shim packs, so readings are less influenced by a single high spot. On a long bed with a casting bow, that averaging effect is worth the extra length.
How it handles the over-sensitivity problem
Two tenths per ten inches is genuinely at the edge of daily usability. Reviewers confirm the accuracy is there, and also confirm the level is heavier and fussier than a .005 in/ft body. Budget time for a settling period on every reading.
If your workflow is repeated setup work rather than one careful dial-in, a coarser level plus a two-collar test will get you to the same place faster and with far less frustration.
5. Digi-Pas DWL1500XY 2-Axis Smart Master Precision Level, Bluetooth
Digi-Pas 2-Axis Smart Master Precision Level DWL1500XY Bluetooth, 0.0002″/ft (0.02 mm/M), Black
0.0002 in/ft resolution
0.02 mm/m accuracy from 0 to 0.5 deg
Bluetooth to iOS and Android
1.3 lb aluminum housing
Pros
- Shows both axes live while you adjust footings
- One person can align a machine instead of two
- Repeatability checks out against granite squares
- Calibration certificate traceable to NIST JIS and DIN
- Solid housing with on-device display
Cons
- On-device LED shows one axis at a time in degrees only
- Phone display can blink off over Bluetooth
- Plastic body has cracked for some owners after a drop
- Extra app licences needed for more than one phone
This is the modern answer to the classic two-person problem. On a traditional setup, one person holds the level and the other walks around adjusting footings. This one streams both axis readings to a phone, so you can watch the readout and turn the adjustment at the same time.
Specifications are 0.0002 in/ft resolution, equal to 0.02 mm/m or 5 seconds, with 0.0004 in/ft accuracy from 0 to 0.500 degrees. Measurement range runs from 0.4 in or 0 degrees to 2.000 degrees, and the calibration certificate is traceable to NIST, JIS and DIN.
Users in machine shops and field service report a substantial cut in alignment time, which matches the obvious advantage: no walking back to the level, no holding a bubble while you crouch somewhere out of sight. Repeatability against granite squares is reported as sound.
The catch is that the dual-axis view lives in the app. The on-device LED shows one axis at a time and reads in degrees only, so if you want both axes at once you need the paired phone or tablet. Bluetooth range is about 50 ft, and the operating range is 10 to 120 degrees F.
Deciding between the two Digi-Pas models
Resolution decides it. This model reads 0.0002 in/ft, while the DWL1300XY later in this roundup reads 0.002 in/ft, a factor of ten coarser. Fine setup and surface work justify the finer model. Rough to medium leveling does not need it.
Both share the same 50 ft Bluetooth pairing, the same temperature range and the same traceable certificate, so the decision comes down to how fine your final pass has to be rather than any other feature.
What one person can and cannot do with it
Owners report it replaces a pair of spirit vial levels for two-axis work, which is a real saving in a cramped machine footprint. It also removes the transcription step, where a helper reads a graduation out loud across a noisy shop.
It does not remove the need for a stable surface. A couple of owners report arriving out of calibration or with base pads showing minimal contact, so verify it on a known flat before a setup day.
6. findmall 8 Inch Level, 0.0002 in/10 in Accuracy
findmall 8 Inch Professional Accuracy Level with Box, 0.0002/10 Inch Accuracy Strip Level for Checking the Work of Machinists Toolmakers
0.0002 in/10 in accuracy, 0.02 mm/m
8 in listing size, carbon steel
4.1 lb
Plastic storage box
Pros
- Performs to stated spec once calibrated
- Good accuracy for lathe and mill leveling at the entry level
- Large bubble that is easy to see
- Adjustable vial can recentre the bubble
Cons
- Calibration screw is coarse and a bump moves the bubble several marks
- Zeroing takes roughly 15 minutes plus re-leveling
- Plastic case is basic and sometimes arrives damaged
- Some units cannot be calibrated because the screw is too sensitive
This is the value entry point, and owners generally agree it does what it says once it has been set. The 8 inch length is the useful one here, since 6, 8 and 12 inch versions are offered and the size decides which machine beds you can span in one pass.
Repeatability has been checked by owners against other precision levels they own, and reviewers describe good accuracy for leveling lathes, mills and pool tables at this tier. The large high-sensitivity bubble is easy to read in workshop lighting.
The single most common complaint is the calibration screw. Owners describe it as coarse and touchy, where a slight bump moves the bubble several marks, and one or two report being unable to zero the level at all because the adjustment is too fine to control.
Budget for that. Owners put the zeroing process at roughly 15 minutes followed by a re-level, which is a one-time cost rather than a daily one, but it does mean you should calibrate it on a bench rather than in the middle of a setup.
How to check accuracy before you trust it
Lay it on a known flat reference and read. Then flip it end for end on the same surface. If the two readings differ by more than a graduation, the level needs calibration and the difference tells you by how much.
That flip test takes ten seconds and can save a day of chasing a reading that was never right. Repeat it on the machine bed if you can reach a long section you already trust.
Whether the storage case matters to you
The box is described as basic, and reviewers report cases arriving damaged. Some listings show a wooden case that does not match what arrives, so treat the case as incidental rather than part of the specification.
Keep the level somewhere dry instead. Carbon steel bodies need it, and a level stored damp will give you a rust problem before it gives you a calibration problem.
7. Starrett 134 Cross Test Level and Plumb, 24-30 Minutes Accuracy
Starrett Cross Test Level and Plumb, 3 x 0.5 x 0.5″, 24-30 Minutes Accuracy
24 to 30 minutes of accuracy
Cast iron with satin nickel finish
3 in long, 0.5 in wide, 0.5 in high
Two vials at right angles
Pros
- Two vials read both axes without repositioning
- Very light and compact for cast iron
- Starrett durability and finish
- Handles plumb square and level in one tool
Cons
- Sensitivity is coarser than a dedicated machinist level
- Small format limits it to short surfaces
The Starrett 134 is a torpedo-class tool with a machinist mindset. It carries two vials mounted at right angles, so it can check two axes at once without being repositioned, and it also gives plumb and square from the same body. The stated accuracy is 24 to 30 minutes of arc, which is far coarser than a machinist level.
That figure is worth converting. Thirty minutes of arc is about .0024 in/ft, or roughly 0.2 mm/m, which places it in the same band as a .002 in/ft machinist level rather than the .005 in/ft band.
Owners value it for exactly what it is: a light, compact cast iron tool for squaring, plumbing and short-range setup work. The satin nickel finish resists corrosion, and at 3 inches long it fits in a pocket without the awkwardness of a full level.
The trade-off owners accept is sensitivity. If your task is bed twist over 40 inches, this cannot help you, because the span is wrong. If your task is checking that a fixture stands square, it is more than good enough.
Where a torpedo level beats a machinist level
Anywhere the measurement is short. Squaring a plate, checking a headstock face, verifying a riser is plumb. Over a 3 inch span, the coarser sensitivity costs you almost nothing in practical terms.
It also wins on handling. Something you can hold in one hand while your other hand holds a scribe is worth more than an extra fraction of accuracy you cannot keep still.
Whether 24 to 30 minutes is enough
For short-range setup it is enough. Converted, it is close to a .002 in/ft machinist level, so you are not working blind, you are simply not working fine.
If your machine has a way where a small error matters over a long reach, use this as a rough check and reach for a proper machinist level for the final pass.
8. Accusize S908-C612 12 Inch Master Precision Level in Fitted Box
Accusize Industrial Tools 12 inch Professional Master Precision Level in Fitted Box, Accuracy 0.0002”/10”, S908-C612
0.0002 in/10 in accuracy in 12 inch
12 in L x 1.83 in W
Hardened steel frame
Adjustable main vial plus cross test vial
Pros
- Adjustable main vial for critical setups
- V-groove base holds position on cylinders
- Cross test vial adds a second direction
- Insulating grip limits thermal drift
- Supplied in a fitted box
Cons
- Higher cost than comparable machinist levels
- Some ratings suggest accuracy or finish falls short
This is the twin of the 12 inch Accusize earlier in the list, and the differences are specific rather than cosmetic. This one carries both an adjustable main vial and a cross test vial, plus an insulating grip, and it comes in a fitted box. Dimensions are given as 12 in L by 1.83 in W, with a body width of 1.574 in and height of 1.890 in.
Sensitivity is again .0002 in/10 in, or 0.02 mm/m. Owners using it for lathe and mill setup appreciate having the cross test vial, because it lets them read a second direction without turning the tool and disturbing the first reading.
The V-groove base is the reason to pick this one over the plainer body. When you are working against a cylindrical surface or a shaft, a flat-only base rocks, and a rocking level produces a reading that looks plausible and is wrong. Owners report the groove holds position.
Ratings sit slightly below its sibling and the review sample is smaller, with limited commentary. Some star ratings suggest accuracy or finish falls short on certain pieces, so check the vial and base flatness on a surface plate before a setup day rather than after.
Two vials versus one
A cross test vial runs at 90 degrees to the main vial. It catches the error you get from canting the level sideways, which is exactly the mistake made when setting a level on a narrow way or a round surface.
That is why it is common on Starrett and Accusize bodies and rare on generic copies. If your machine has both flat ways and a V-way or round reference, a two-vial body saves you a second pass.
Where the fitted box earns its place
A level this sensitive needs protecting between uses. The box keeps the ground base off the bench and stops the vial from taking a knock, which is the most common way a good level quietly becomes an uncalibrated one.
Owners who store levels loose report rust and damaged vials over time. Treat the box as part of the tool rather than as packaging.
9. Digi-Pas DWL1300XY 2-Axis Smart Machinist Digital Level, 0.002 in/ft
Digi-Pas 2-Axis Smart Machinist Digital Level DWL1300XY Bluetooth 0.002″/ft (0.2mm/M), Black
0.002 in/ft resolution, 0.18 mm/M
0.004 in/ft accuracy to 1 degree
Measurement range to 5 degrees
0.45 kg zinc alloy body
Pros
- Two-axis digital readout replaces a pair of vial levels
- Wireless phone display enables one-person alignment
- Cuts alignment time over back and forth bubble checking
- Lighter zinc body is easy to reposition
- Includes app card
- quick guide and calibration certificate
Cons
- Lower sensitivity than the DWL1500XY so it suits rougher setups
- Dual-axis display depends on the paired phone
- App must be downloaded and unlocked before use
This is the coarser of the two digital machinist levels here, and it is deliberately so. Resolution is 0.002 in/ft, equal to 0.18 mm/M or 40 seconds of arc, with stated accuracy of 0.004 in/ft or 80 seconds from 0 to 1.00 degrees. Measurement range reaches 5.000 degrees, five times the reach of the finer model.
That combination makes it the right digital tool for a first alignment pass, where you are bringing a machine from nowhere to roughly level before the precision stage begins.
Users deploy it mostly for rough to medium machine leveling and report that the faster one-person setup is the main win. The zinc alloy body is lighter than the housing on the finer model, which matters when you are repositioning it repeatedly on a machine base.
Reviewers are straightforward about the limit: it is accurate enough for the job, but a finer resolution model is preferable for precision work. The app must be downloaded and unlocked from the included redemption card before the level is fully usable.
Resolution versus accuracy, explained
Resolution is the smallest change the readout will show. Accuracy is how much you can trust it. Here resolution is 0.002 in/ft while accuracy is 0.004 in/ft, roughly double, because no digital level holds its stated accuracy across the whole range.
Read the accuracy figure when you care about the number you are about to cut to. Read the resolution figure when you care about seeing movement at all.
How this fits a two-stage setup
Use this model to get close fast, then finish with a bubble machinist level or a two-collar test. Owners report it saves real time on the coarse stage because one operator can watch the readout and adjust at the same time.
The wider 5 degree measurement range is useful when a machine is badly out, since readings do not drop off the scale as quickly as they do on a finer model.
10. WFLNHB 12 Inch Level, 0.0002 in per 10 inch
12 Inch Level Accuracy 0.0002 Per 10 Inch Machinist Tool Used to Test the Horizontal Position of Equipment and Flatness
0.0002 in per 10 in sensitivity
12 in listing size
6.1 lb carbon steel body
Shifts in the highest sensitivity class
Pros
- Advertised sensitivity at the entry level
- Carbon steel body feels substantial on a bed
- Box keeps it stowed in a toolbox
- 12 inch span covers wider way spacing
Cons
- Smallest review base in this roundup at 36 ratings
- Bubble must be fully settled before a reading counts
- Lowest average rating here at 4.0
This is a 12 inch carbon steel machinist level listed at .0002 in per 10 in, which is 0.02 mm/m, and it is aimed at checking guideway straightness and installed horizontal or vertical position. The listing gives 12 in packaging dimensions and a 6.1 lb weight.
Owners like the solid feel of the carbon steel on a machine bed and consider the stated sensitivity a good result at this tier. The included plastic box is a bonus for toolbox storage rather than bench storage.
The honest caveat is the review base. At 36 ratings it has the smallest sample in this roundup, and its 4.0 average is the lowest, which usually points to some inconsistency in finish or accuracy against the listing claims rather than a design fault.
The listing also states that readings must be taken after the bubble becomes fully stationary. On a level this sensitive that wait is longer than you would like, and it is a reminder that settling time is a real cost of the finest class.
Why the 12 inch span matters on a wide machine
If your way spacing is wider than 8 inches, an 8 inch level has to be repositioned and overlapped. Overlaps are where error accumulates, and a 12 inch body removes them for most setups short of a full-size VMC.
For a wide VMC it is the frame width that actually limits you, not just the length, so measure the clear span before committing to a long level.
How to buy this one without regret
Buy it expecting to calibrate and re-check it, and treat the flip test as mandatory on arrival. With a small review base, your own verification matters more here than for a level with hundreds of ratings behind it.
Keep the plastic box and store it dry. Carbon steel with a fine vial is not a tool to leave out in a damp shop between setups.
Bubble vs Digital vs Laser Levels: Which One Do You Need?
A machinist level with a graduated vial gives you a numeric reading directly in inches per foot, in the tool itself, with no power and nothing to pair. That makes it the default choice for bed twist work, because the number is the answer.
A digital machinist level reads both axes at once and streams to a phone. You trade the direct read for speed and gain one-person operation. You also gain an app dependency, a battery and a calibration certificate to keep track of.
A laser level projects a plane or a line. It is excellent for rough levelling over a large area and for setting out columns and frames, but it is a poor tool for reading a small twist in thousandths across a machine way.
For machine setup specifically the ranking is graduated vial first, digital second, laser for the rough pass only. Plenty of shops own all three and use them in exactly that order.
How to Choose the Right Level Length
Measure the distance between the points you need to compare before you buy anything. A level wider than your machine’s clear span is wasted money, and one shorter than the span forces overlapping readings that add error.
As a working rule, 6 inch covers most small lathes and hobby mills where you are comparing points within a short distance. 8 inch covers standard bed widths and is the most commonly useful single length. 12 inch suits wide VMCs and large mills where the way spacing exceeds what a shorter body can bridge in one pass.
Also check the frame width, not just the length. The body has to sit flat on the ways without overhanging a V-way, and a wide frame on a narrow way is how readings go wrong quietly.
How to Level a Lathe or Mill With a Precision Level
Start coarse. Get the machine roughly level on its leveling pads with an ordinary level, tightening anchor bolts evenly. Nothing fine is worth doing until the base is stable and the feet are not springing.
Step two is the twist check. Place the precision level on the ways near the headstock and read the graduation. Move it to the tailstock end and read again. If the two readings match, the bed is untwisted.
Step three is the shim calculation. Work out the difference between the two readings, then multiply by the span between your measuring points. That gives you the thickness to add or remove at the low corner. A coarser level makes this easier than squinting at a fine vial.
Step four is to unbend before you align. If a precision straight edge shows a bow along the bed, straighten that first. Then align the tailstock to the bed, and only then the headstock to the tailstock.
Finally, re-check on a schedule rather than chasing perfection in one session. Residual stress in a casting relaxes over days, so check daily until the readings stop moving, then weekly, then monthly, then annually.
Calibration, Vial Replacement and the Vials-Only Upgrade Path
To calibrate a precision level: place it on a verified flat surface, loosen the vial retaining nuts, set the bubble between the marked lines, reverse the level end for end, and if the reading differs, split the difference between the two positions. Re-tighten and verify again.
That is a bench job, not a setup-day job. If you never reverse the level during a single setup, calibration matters less than most owners assume.
The upgrade path worth knowing is the vial-only swap. A coarser body fitted with a finer vial does the job of a finer level, provided the base is flat. One owner described keeping a coarser body, installing a more sensitive vial and scraping the base to improve flatness. Scraping is a real technique for improving contact, but it is not something to do casually.
A worn or cracked vial can often be replaced without buying a new level, and a used body from an earlier model is a legitimate way into the class. What you should not assume is that a used level arrives calibrated. Treat any used level as a body of unknown condition with a calibration job ahead of you.
Why Over-Sensitive Levels Are a Disadvantage
Beyond roughly .0005 in/10 in, sensitivity stops being a feature. The bubble takes longer to settle, thermal drift from your hand moves it, and the reading is hard to hold long enough to write down.
Owners report this plainly. Someone with two Starrett 98 levels and a far more sensitive import reports that the sensitive one is frustrating to use, so it lives in the box. Another first-time buyer spent a similar amount on a mainstream level and felt far more comfortable with it.
There is a second effect people miss. A very fine level makes you chase readings. Because it responds to everything, you keep finding small differences and keep adjusting, and the machine ends up shimmed to a state nobody would have chosen deliberately.
The practical advice from the forums is to treat .005 in/ft and .002 in/ft as the practical ceiling for regular work, keep the finer levels for the final pass, and let the two-collar test handle the last adjustment. Sensitivity you cannot use is not accuracy.
Frequently Asked Questions
How accurate are machinist levels?
Standard machinist levels run .005 in/ft, which is 0.42 mm/m or 80 to 90 seconds of arc. True master precision levels reach .0002 in/10 in, equal to 0.02 mm/m. For regular machine setup, about .0005 in/10 in is the practical ceiling, because finer levels are hard to keep settled long enough to read.
What is a machinist level 3?
A machinist level with a .0002 in/10 in vial, often called two tenths per foot. It is the most sensitive common class and is used for fine dial-in and surface work rather than routine machine leveling. Many owners report it is more sensitive than they can use comfortably.
How accurate are Starrett 98 levels?
The Starrett 98 series is specified at .005 in/ft, equal to 0.42 mm/m or 80 to 90 seconds of arc, with an adjustable ground vial in a knurled brass tube on a precision-machined cast iron base. Check the vial tightness and base flatness on arrival, because some current models are imported rather than made in the USA.
How to calibrate a precision level?
Place the level on a verified flat surface, loosen the vial retaining nuts, set the bubble between the marked lines, reverse the level end for end, then split the difference between the two positions if the readings differ. Re-tighten and verify. A level only needs to be accurate if you reverse it during a setup.
Do I need a machinist level to level a lathe?
A precision level speeds up dial-in considerably, but a two-collar test with a test bar can finish the job without one. Get close with an ordinary level, then use a .0005 in/10 in machinist level or a two-collar test to complete the alignment.
What size machinist level do I need?
Measure the distance between the points you need to compare. Six inch spans most small lathes and mills, eight inch covers standard bed widths, and twelve inch suits wide VMCs and large mills. A level longer than your clear span is wasted, and one shorter forces overlapping readings.
Is a more sensitive precision level better?
Not for machine setup. Beyond about .0005 in/10 in the bubble is slower to settle and more easily moved by hand heat, which makes it harder to use. Owners with very sensitive levels report keeping them in the case rather than the setup box.
Conclusion
For most shops the Starrett 98-6 is the right first precision level for machine setup, because it is accurate enough for real levelling work, adjustable, and it comes with the cross test vial that stops the most common placement error. Buy it, calibrate it once on a flat surface, and use it to compare readings rather than chase a centred bubble.
If your bed is long enough that 6 inches will not span it, step up to the Accusize 12 inch. If you would rather work alone while reading both axes, the Digi-Pas DWL1500XY covers that, with the coarser DWL1300XY for the rough pass. Keep the very sensitive levels for final work, and let the two-collar test take care of the last adjustment.
Whichever you choose, measure the span before you buy, check the level on a known flat when it arrives, and re-check the machine on a schedule instead of chasing perfection in one afternoon.