
Dimensional Tolerances in CNC Prototypes: What to Specify and What Not to Over-Tighten
Specifying tighter tolerances than the function requires does not improve your prototype — it makes it more expensive and slower to deliver. This guide tells you which dimensions actually need strict tolerances and which ones you can leave to standard.
The most expensive tolerance you can put on a drawing is the one you don't need.
In CNC prototypes, over-specification — assigning tighter tolerances than the function requires — is the silent mistake that doubles quotes, extends lead times, and does nothing to improve part performance. The opposite error, leaving critical dimensions unspecified, is equally costly: the shop applies its general tolerance and the assembly fails on the first test.
This guide is for engineers who want to calibrate correctly: which dimensions need tight tolerances, which can go to standard, and when GD&T adds value versus when it just adds cost.
In Summary
- Standard CNC machining tolerance: ±0.1 mm without additional specification; ±0.05 mm for precision dimensions; ±0.025 mm or less for high-demand zones
- Cost vs. tolerance is not linear: going from ±0.1 mm to ±0.01 mm can multiply part cost by 3 to 5x
- Only 20 to 30% of dimensions in a typical prototype are functional — the rest can go to standard tolerance without affecting the outcome
- GD&T is not a default for prototypes: use it where position or geometric form is the function, not as a general precaution
- More on prototype process selection: From CAD to Prototype in Days
The problem is not that engineers don't know what a tolerance is. The problem is that when there is uncertainty about whether something will work, the natural reaction is to tighten everything. "If I put ±0.05 mm on all dimensions, at least the machining won't be the failure." That reasoning makes sense in production. In prototypes, it costs time and money you don't have.
1. The Tolerance Hierarchy in CNC Prototypes
Before specifying any tolerance, classify your dimensions into three categories:
Critical functional dimensions
These are the ones that determine whether the part works: fits between mating parts, shaft diameters that go into bearings, sealing surfaces, lengths that define interference. Tight tolerances apply here: ±0.025 mm or less depending on the fit required.
For standard shaft-hole fits, use ISO 286 nomenclature. An H7/f7 sliding fit gives you controlled clearance without inventing custom dimensions. It is clearer for the machinist and easier for you to verify.
Geometric reference dimensions
Center-to-center distances between bores, flatness of a mounting surface, perpendicularity of a shaft. These need tolerances, but rarely the tightest ones. ±0.05 mm to ±0.1 mm is sufficient for most validation scenarios.
GD&T comes in here when form or position is the function — a bore that needs to be within 0.2 mm of true position relative to a datum face — not as decoration on the drawing.
Non-functional dimensions
Reference lengths, non-critical corner radii, pocket depths with no contact with another part. These go to shop standard tolerance (±0.1 mm) and should stay there.
The concrete exercise: go through your model dimension by dimension and ask "if this is 0.1 mm off, does the prototype fail at its purpose in this iteration?" If the answer is no, don't add a tight tolerance.
2. The Real Cost of Tightening Tolerances
Machining cost does not rise linearly with tighter tolerance. It rises in steps, and the jumps are significant:
±0.1 mm — standard tolerance for conventional CNC milling and turning. No extra steps required. The operator measures with a micrometer, adjusts the offset, continues.
±0.05 mm — within the standard machining range at a good shop. May require a dedicated finishing tool and an extra pass. Moderate cost increase.
±0.025 mm — this is where thermal stability starts to matter. The shop needs to control temperature, use quality gauge tips, and reduce tool feed rate. Cycle time increases.
±0.01 mm or less — we are in grinding or honing territory, not conventional milling. This requires an additional process, CMM measurement on every part, and sometimes multiple adjustment iterations. Cost can be 5x that of ±0.1 mm on the same geometry.
For a form-and-fit validation prototype, most dimensions you are tolerancing at ±0.05 mm could go to ±0.1 mm with no consequence. That change can reduce the quote by 20 to 30% and cut lead time by a full day.

3. What You Always Need to Specify
There are dimensions that engineers frequently leave unspecified and that cause assembly problems:
Functional threads. If the thread will carry load or seal fluid, specify the fit class (6H/6g for standard metric threads). An M8 thread without a class can come out loose or with a forced entry depending on the shop.
Surface finish in contact zones. Ra 1.6 µm is the machining standard. For O-ring sealing surfaces you need Ra 0.8 µm or better. For bearing seats, Ra 0.4 µm. If you don't ask for it, the shop delivers what comes off the process without an extra step.
Bore diameters for fits. A 12 mm bore for a 6001 bearing needs H6 tolerance (12.000–12.011 mm). Without that specification, ±0.1 mm is valid — and that is too much clearance for a bearing fit.
Parallelism and flatness on mounting surfaces. If the part will bolt to another or serve as a measurement datum, a flatness tolerance of 0.05 mm prevents alignment problems that are difficult to diagnose later.
The technical reference for this is ASME Y14.5-2018 for GD&T and ISO 286 for dimensional fits.
4. When GD&T Is Worth It on a Prototype
GD&T makes sense when a linear dimension cannot capture the functional requirement. Specific cases:
- True position of bores in a pattern — if you have 6 bores that need to be within 0.3 mm of true position for a cover to close, a linear location tolerance does not guarantee the same result
- Concentricity of diameters — a shaft with two diameters that must be concentric (toolholder, bearing sleeve)
- Flatness as a measurement datum — if you will measure the prototype with CMM and need a reliable reference surface
What does not need GD&T on a prototype:
- Corner radii with no functional role
- Entry chamfers
- Overall reference lengths
- Pocket position with no interface to another part
Adding GD&T controls to non-functional dimensions does not make the drawing more rigorous — it makes the machinist take longer to interpret and measure, and you pay for that time.
5. How to Present the Drawing to Your Supplier
A useful prototype drawing has:
- A general tolerance block in the title block: "General tolerance: ±0.1 mm, angles ±0.5°." This frees the machinist from guessing on non-critical dimensions.
- Specific callouts only where they matter — clearly marked, without redundancy.
- Surface finish specified in contact zones, not as a global symbol across the entire drawing.
- Material and treatment note if applicable — not all steels machine the same way, and heat treatment affects final dimensions.
What clutters a drawing without adding value: tolerances equal to the general tolerance repeated as individual dimensions, contradictory notes, GD&T blocks copied from production drawings without adaptation to the prototype context.
If you don't have the drawing ready and need to evaluate whether the geometry is viable before investing in the full drawing, Radii's FeasibilityAI can review the 3D model and flag high-complexity zones — avoiding surprises in the quote.
See also: 5-Axis CNC Machining for R&D Prototypes and From CAD to Prototype in Days

Frequently Asked Questions
What is the standard tolerance in CNC machining without additional specification?
Most shops apply ±0.1 mm as a general tolerance when no specification is given. For critical dimensions — fits, threads, sealing surfaces — you must always specify explicitly on the drawing.
When do I need to apply GD&T on a prototype?
GD&T is necessary when there are requirements for position, perpendicularity, concentricity, or flatness that a simple linear tolerance cannot control. For basic functional validation, linear dimensions on critical zones are often sufficient.
Why can a ±0.01 mm tolerance cost five times more than ±0.05 mm?
At tolerances below ±0.025 mm, the machinist needs to control shop temperature, use dedicated finishing tools, measure every part with CMM, and reduce cutting speeds. That extra time translates directly into cost.
What happens if I don't specify tolerances on my prototype?
The shop applies its general tolerance (typically ±0.1 mm). If your prototype has fits, functional threads, or sealing surfaces, those dimensions may fall outside what you need without anyone knowing until assembly.
Conclusion: Tight Where You Need It, Standard Where You Don't
- Classify your dimensions: critical functional, geometric reference, and non-functional
- Apply ±0.025 mm or less only where the function demands it — cost rises exponentially
- Always specify: functional threads, surface finish in contact zones, and bore diameters with ISO fit class
- GD&T on prototypes: only when a linear dimension cannot capture the real requirement
- A general tolerance block in the title block simplifies the drawing and reduces misunderstandings
Radii connects your prototype with audited CNC shops in Mexico that read technical drawings, not just STEP files. Upload your model at app.radii.com.mx and get a quote with a feasibility review before finalizing the drawing.