3D Printing Tolerances and Fit
Nothing fits at nominal size. Here are the clearances that make printed parts assemble properly, and why a test print beats guessing.
The most common reason a printed part does not fit is that it was drawn at the exact size of the thing it fits. That is a machinist's instinct and it does not survive contact with FDM. Printed plastic has a mind of its own, and you design around it rather than against it.
What accuracy to actually expect
For a well-set-up FDM machine printing a typical part, think in terms of roughly ±0.2 mm on a small dimension, drifting to something like ±0.5 mm or a bit more across a large part. That is a realistic working figure, not a specification — it moves with material, part size, geometry, temperature and print speed.
Compare that with machining, which routinely works an order of magnitude tighter, and you can see where FDM sits. It is fine for the overwhelming majority of functional parts. It is not fine when you need precision, and pretending otherwise wastes your money.
Why parts come out the wrong size
Shrinkage. Plastic is extruded hot and contracts as it cools. Every material does it, by a different amount — PLA barely, PETG a little more, ABS and ASA noticeably more. On a 100 mm ABS part that adds up to something you can measure with a ruler.
Squashed layers. The first layer is deliberately pressed into the bed, so it spreads. That gives the "elephant's foot" — a slight bulge around the bottom that stops a part sitting flat or entering a pocket. A small chamfer on the bottom edge fixes it.
Holes come out small. Two reasons: the plastic pulls inward as it cools, and on the inside of a curve the beads are squeezed together. A 5 mm hole commonly measures around 4.8 mm. This catches everyone.
Outside dimensions come out big. The mirror image of the same effect — the outer bead bulges slightly outward.
So as a rule of thumb: holes shrink, posts grow. Both errors work against you when you try to fit one into the other.
Clearances that actually work
These are starting points, in PETG or PLA, for parts printed on the same machine:
- Press fit / interference (assembled with force, stays put): 0.1 mm total clearance, or slightly negative.
- Snug fit (goes together by hand, no play): 0.2 mm.
- Sliding fit (moves freely, minimal slop): 0.3 to 0.4 mm.
- Free / loose fit (definitely no binding, hinges, lids): 0.5 mm and up.
- Threaded rod or bolt through a clearance hole: 0.5 mm on diameter, minimum.
"Clearance" here means total diametral clearance — split it between the two parts, or take it all off one.
Two things to note. First, these are for printed part to printed part. If one side is a bought component — a bearing, a bolt, a magnet, a piece of tube — you have less to play with, because the bought part is right and the printed part has to do all the accommodating. Second, ABS and ASA need more, because they shrink more.
Fitting to something that already exists
This is the most common real job: a bracket around an existing tube, a pocket for a specific device, a clip into a specific hole.
- Measure with calipers, not a ruler. A ruler is fine for envelope sizes and hopeless for a fit.
- Measure the same thing more than once, in more than one place. Extruded and moulded parts are rarely as round or as square as you think.
- Tell us which dimensions are critical and which have room to move. This is the single most useful thing you can send us — it lets us put the error where it does no harm.
- Send the part if you can. Posting us the thing itself beats any set of measurements.
Our guide on measuring a bracket covers the method.
Test prints are cheap; reprints are not
If a fit is critical, print the fit — not the part. A small test coupon with just the hole, the boss or the pocket takes minutes and a few grams, and it tells you exactly what clearance works for that material on that machine. Getting the number right on a small test and then printing the real part once is faster and cheaper than printing a large part three times.
For anything fiddly, we would rather run a small test than have you pay for a big part that does not go together. Say so at the quote stage.
Design around the problem, not into it
- Drill critical holes rather than printing them to size. A printed pilot hole and a drill bit gets you machining accuracy in seconds.
- Chamfer every lead-in. A tapered entry makes an assembly forgiving of a tenth of a millimetre.
- Use a slot instead of a hole where a fixing does not need to be exact — it absorbs error.
- Design in adjustment. A screw and a slot beats a fit that has to be perfect.
- Split the difference. Where two printed parts mate, take clearance off both.
More of this in our design tips.
When FDM is the wrong process
Straight answer: if your part genuinely needs tolerances tighter than about ±0.1 mm, or a real bearing fit, or a precision thread, FDM is not the right tool. Nor is it right for a precision inspection gauge — printed plastic moves with temperature and humidity too much to be a reference. Those are machining jobs, and we would rather say so than take the order and hand you a part that does not work.
For everything short of that — which is most parts — designing in the right clearance is all it takes.
Get an estimate · upload a file · see the design help service.
Models that show this in practice
Open-source designs from our print library. Each one has a full material and quantity price breakdown.
Gusseted Shelf Bracket (large)
Control Knob (large)
Gusseted Shelf Bracket (medium)
Gusseted Shelf Bracket (small)
These are open-source example designs (CC0) we publish to show what the process suits and what it costs — not a record of past jobs. Prices shown are examples in PLA.
Get a 3D print estimate
Upload your file or describe the part. We review printability before confirming anything.
Services covering this
Related guides
Prototyping a Product: A Step-by-Step Guide
The route from a sketch to a batch you can sell — what each stage is for, what to test, and when to stop printing in PLA and start printing in something real.
How Many Prototype Iterations Should You Expect?
Three or four rounds for a simple part, more for anything with a mechanism. How to iterate cheaply — one change at a time, several variants per print — and how to know when to stop.
From Prototype to Production
What happens after the design is frozen — bridge batches, the point where tooling beats printing, the design changes that make moulding cheaper, and what a moulder actually needs from you.