Tolerance in 3D printing is a more nuanced topic than machining. Instead of a single number (± 0.025mm on a mill), 3D printing tolerances depend on the process, the material, the part size, the feature orientation, and sometimes even the temperature of the room. Understanding how each process actually behaves — rather than what the spec sheet promises — is the difference between prototypes that mate with existing hardware on the first try and prototypes that need a file and sandpaper.
This guide walks through what to expect from each technology, where the variation comes from, and how to design around it.
What tolerance numbers actually mean
When a 3D printing service quotes "±0.2mm tolerance," that number is typically the best-case deviation for a feature of a given size — often a 10mm nominal dimension in the X-Y plane. Actual tolerance varies with:
- Part size. Larger parts have more absolute variation. A 200mm part rarely holds ±0.2mm end-to-end.
- Feature orientation. Vertical dimensions (along the build axis) are typically looser than horizontal dimensions.
- Material. Semi-crystalline materials (Nylon PA12) shrink more on cool-down than amorphous resins.
- Geometry. Tall thin walls deform more than short thick ones.
- Post-processing. Sanded, painted, or vapor-smoothed parts have different tolerances than as-built.
Tolerance by technology
FDM
Typical: ±0.2mm on small features, ±0.5mm on 200mm+ dimensions. FDM tolerance is dominated by layer bonding and thermal contraction. Features in the X-Y plane hold tighter than features along Z. Horizontal holes print smaller than nominal because of overhang droop — oversize hole diameters by 0.2–0.4mm if they need to match a fastener.SLA
Typical: ±0.1mm on small features, ±0.2mm on 100mm+ dimensions. SLA is the tightest of the three common processes because the laser spot is small and the resin cures in place with minimal thermal movement. Mating features for snap-fits are reliable directly off the machine.Industrial SLA
Typical: ±0.2mm on small features, ±0.5mm on 500mm+ dimensions. Large-format SLA holds roughly the same relative accuracy as desktop SLA but the absolute numbers grow with part size.MJF
Typical: ±0.3mm on small features, ±0.5mm on 200mm+ dimensions. MJF nylon shrinks about 3% on cool-down, and the HP build algorithm compensates for expected shrinkage — but parts still move from batch to batch based on powder age, build density, and cool cycle.Design rules for mating features
When your 3D printed part has to fit into existing hardware (an injection-molded part, a machined housing, a bearing press-fit), oversize the printed hole and undersize the printed shaft:
- Printed hole to receive a shaft: Oversize by 0.3–0.5mm beyond the nominal shaft diameter.
- Printed shaft to fit into a machined hole: Undersize by 0.2–0.3mm.
- Snap-fit joints: Design with 0.2–0.4mm clearance between mating surfaces.
- Press fits: Do not design 3D printed press fits at all. Use threaded inserts or screw-in fasteners instead.
What to do if tolerance is critical
If your part has a tolerance tighter than ±0.1mm on any feature, 3D printing alone will not reliably hold it. Options:
- Machine the critical feature after printing. Drill holes to final size, mill flat surfaces flat. The print provides the rough stock.
- Use SLA for the whole part. SLA is the tightest 3D printing process and may meet your spec on small parts.
- Switch to CNC. If tolerance is the top priority and the geometry allows it, CNC is the better process.
