Tolerance Calculator

Engineering design is constantly trading manufacturing cost against functional requirements. Tighter tolerances guarantee better fit and performance but cost exponentially more. Loose tolerances are cheap but risk fit and function problems.

The art of design is choosing the loosest tolerance that still meets the functional requirements — and no looser. This calculator helps you check the math both ways.

To calculate a dimensional tolerance, add the upper and lower deviations to the nominal size: Maximum Limit = Nominal + Upper Deviation, and Minimum Limit = Nominal + Lower Deviation. The tolerance width (total allowable range) is then Tolerance = Max Limit − Min Limit, which equals Upper Deviation − Lower Deviation.

For a 25.000 mm shaft with deviations of +0.021 and 0, the max limit is 25.021 mm, the min limit is 25.000 mm, and the tolerance width is 0.021 mm (21 micrometers). The mid limit is (Max + Min) / 2 = 25.0105 mm.

This deviation-based method is the foundation of the ISO 286 limits-and-fits system (Encyclopaedia Britannica).

Tolerances are expressed in the same unit as the nominal dimension, most commonly the millimetre (mm), the SI unit of length. Because tolerances are often very small, engineers frequently work in micrometres (µm), where 1 mm = 1000 µm; a tolerance of 0.021 mm is equivalently 21 µm.

In the United States, tolerances are also stated in inches or in "thou" (thousandths of an inch, or mils), where 0.001 in ≈ 25.4 µm. The metre is the SI base unit of length as defined by the BIPM and NIST.

Always keep both parts of a fit in the same unit system before combining hole and shaft values to avoid conversion errors.

For a mating fit, compute the extreme clearances between the hole and shaft. Maximum Clearance = Hole Max − Shaft Min, and Minimum Clearance = Hole Min − Shaft Max.

The sign of these clearances determines the fit type:

• If Minimum Clearance is positive, the shaft is always smaller than the hole, giving a clearance fit.
• If Maximum Clearance is negative, the shaft is always larger, giving an interference (press) fit.
• If the range spans zero — min clearance negative but max clearance positive — the result is a transition fit.

For example, a hole 25.000/25.021 mm with a shaft 24.980/24.993 mm gives max clearance 25.021 − 24.980 = 0.041 mm and min clearance 25.000 − 24.993 = 0.007 mm, a clearance fit.

ISO 286 is the international standard governing limits and fits for cylindrical features. It defines 20 standard tolerance grades, from IT01 and IT0 through IT1 to IT18, where lower numbers mean tighter tolerances.

Each grade sets the tolerance width as a function of nominal size, so a given IT grade produces a larger physical tolerance on a large part than on a small one. Fundamental deviations, labelled with letters (lowercase a–zc for shafts, uppercase A–ZC for holes), fix the position of the tolerance band relative to the zero line.

Combining a grade and a deviation letter — such as H7/g6 — fully specifies both parts of a fit in compact notation (Encyclopaedia Britannica).

In the hole-basis system, the hole is held at H (lower deviation zero) and the shaft varies. The main fit categories are:

• H7/g6 is a close running or free-running fit: the shaft is slightly undersized, producing a small guaranteed clearance for lubricated sliding or rotating components.
• H7/k6 and H7/n6 are transition fits used for accurate location where parts must be assembled and disassembled with light force.
• H7/p6, H7/s6, and tighter combinations are interference or press fits that require force, heating, or cooling to assemble and rely on residual stress to transmit load.

Selecting the correct fit balances assembly method, load transfer, and the ability to service the joint later.

Tolerance analysis appears wherever parts must mate reliably. Common examples span many industries:

• Rolling-element bearings use interference fits on the rotating ring and clearance or transition fits on the stationary ring so that shafts run true without spinning in their seats.
• Hydraulic pistons and cylinders rely on tight clearance fits to seal pressure while still sliding.
• Precision instruments and optics keep alignment within microns using tight IT grades.
• Automotive and aerospace assemblies use tolerance stack-up analysis to guarantee that dozens of stacked parts still fit in the worst case.
• Even 3D-printed and injection-moulded parts specify tolerances to account for shrinkage.

Choosing tolerances that match the real function — not tighter — is central to design for manufacturability.

The achievable tolerance depends directly on the manufacturing process. Different processes hold different grade ranges:

• Precision grinding, honing, and lapping reach the tightest grades (roughly IT4–IT6), producing tolerances of a few micrometres to about 15 µm on small parts.
• Turning and milling typically hold IT7–IT10, corresponding to tolerances from tens of micrometres to a few tenths of a millimetre.
• Drilling, casting, and rough machining fall in the coarser IT11–IT16 range.

Because tolerance width scales with nominal size in ISO 286, the same IT grade is harder to hold on a large workpiece. Matching the required IT grade to a process the shop can reliably deliver avoids scrap, rework, and inspection overhead.

Tolerance stack-up predicts how individual part variations accumulate across an assembly. Two main methods are used:

• Worst-case analysis, the simplest method, sums all the individual tolerances arithmetically: the maximum possible gap or interference equals the sum of every contributing tolerance in the chain. This guarantees fit but is conservative and can force needlessly tight parts.
• Statistical stack-up (root-sum-square, RSS) instead combines tolerances as the square root of the sum of their squares, recognising that all parts rarely reach their extremes simultaneously. RSS predicts a tighter, more realistic assembly variation for high-volume production.

Engineers pick worst-case for safety-critical or low-volume joints and statistical methods where production data supports the assumption of independent, centred distributions.

The most common and costly mistake is over-tightening tolerances beyond functional need, since cost rises steeply as tolerances shrink. Other frequent errors include:

• Mixing unit systems (mm versus inches) when combining hole and shaft values.
• Forgetting that a fit requires both parts' tolerances rather than one.
• Confusing bilateral (±) with unilateral (one-sided) deviation callouts.
• Misreading the sign convention: a negative lower deviation lowers the minimum limit, while an unstated deviation defaults to zero.
• Applying the same IT grade to parts of very different sizes without checking the resulting physical tolerance.
• Ignoring geometric tolerances (form, orientation, position), which can allow parts that pass size checks yet still fail to assemble.

Nominal tolerances assume a reference temperature of 20 °C (68 °F), the standard adopted by ISO and NIST for dimensional metrology.

Because metals expand when heated, a part measured warm reads larger than at 20 °C; steel expands roughly 12 micrometres per metre per degree Celsius, so a 100 mm steel part gains about 1.2 µm per °C — enough to swallow a tight tolerance.

Surface roughness also matters: measured size depends on where a probe or gauge contacts peaks and valleys, so a rough finish blurs the effective dimension.

For critical fits, engineers control both temperature during inspection and specify surface finish alongside size so that the calculated limits reflect real assembled behaviour.