What Are Realistic Tolerances in 3D Printing, How Can You Achieve Them?

Why Are 3D Printing Tolerances Inherently Different from CNC?

Unlike subtractive machining, where a solid block is carved to a precise dimension, 3D printing is an additive, thermal, and layered process. This creates unique sources of variation:

• Thermal Dynamics: Materials shrink as they cool (sintering) or cure (polymerization). This shrinkage is often non-uniform.
• Layered Anisotropy: The Z-axis (build direction) tolerance is fundamentally different from X-Y axis tolerance. Layer adhesion and stair-stepping on curves are Z-axis challenges.
• Toolpath Dependency: The path the nozzle or laser takes can affect dimensions. Corners can bulge due to inertia, and holes can become elliptical due to path rounding.

Consequently, quoting a single “±X mm” tolerance is misleading. You must think in terms of a tolerance budget that accounts for multiple, interacting variables.

What Are the Primary Sources of Dimensional Variation?

Your tolerance budget is spent in four key areas:

1. How Does Material Behavior Dictate Shrinkage and Warpage?

This is the largest, most predictable variable. All materials change volume during processing.

• Semi-Crystalline Polymers (Nylon, PEEK): Exhibit significant, anisotropic shrinkage (1.5-3.0%) as molecules crystallize upon cooling, leading to pronounced warping. This must be compensated for in the slicer or CAD model.
• Amorphous Polymers (ABS, PLA, Resins): Shrink less (0.2-1.5%) but are prone to warping from residual stress as they cool unevenly. Enclosures are critical.
• Metals (Stainless Steel, Aluminum, Titanium): Shrinkage is significant (2-4%) but is highly predictable and isotropic when process parameters are locked. This is compensated for via uniform scaling in the build file.

A Real Data Point: A client printing a large PA12 (Nylon) duct saw a 2.1% linear shrinkage after SLS printing and cooling. The 200mm design dimension printed at 196mm. By applying a corrective scaling factor of 102.1% in the build preparation software, they achieved a final part within ±0.2mm of the target.

2. What is the True Impact of Machine Calibration and Repeatability?

A printer’s inherent precision sets the baseline.

• Desktop FDM: Backlash in belts, Z-axis wobble, and inconsistent extrusion can lead to ±0.2-0.5 mm variations even on small parts. Repeatability is the bigger challenge than absolute accuracy.
• Industrial FDM/SLS/SLA: With precision linear rails, closed-loop control, and controlled environments, machine-to-machine and job-to-job repeatability can be held within ±0.1 mm or better.
• Laser/Pixel Size: In SLA/DLP and SLM, the size of the laser spot or LCD pixel defines the minimum feature size and edge definition. A 75μm laser spot cannot produce a sharp 25μm corner.

3. How Do Geometry and Orientation Consume Your Tolerance Budget?

The part’s design and how it’s placed on the build plate are critical.

• The “Aspect Ratio” Problem: A tall, thin feature will have worse Z-axis tolerance than a short, wide one due to layer adhesion variability and potential wobble.
• Overhangs and Supports: Any surface touching a support will have degraded surface finish and dimensional fidelity. A critical bearing surface should be oriented to print support-free.
• Hole Diameter and Pin Size: Holes tend to print smaller (as the toolpath traces the inside), and pins tend to print larger (as the toolpath traces the outside). A standard rule is to undersize holes and oversize pins in your CAD model by 0.2-0.5mm, depending on the process.

4. How Does Post-Processing Change Dimensions?

The part isn’t finished when the print ends.

What Are Achievable Tolerances by Technology?

These are realistic, industrially achievable tolerances for well-controlled processes, not best-case lab numbers. They assume optimal orientation, mature material, and professional post-processing.

Key Insight: Notice the percentage-based specifications on industrial machines. This is correct: absolute error often scales with part size. A 10mm feature might be held to ±0.1mm (1%), but a 200mm feature on the same machine might be held to ±0.3mm (0.15%).

What is the Practical Framework for Designing and Specifying Tolerances?

Adopt this workflow to engineer success from the start.

Step 1: Define Functional vs. Aesthetic Tolerances

Not all dimensions are equal. On your drawing or in your mind, categorize:

• Critical Functional Dimensions (CFDs): Surfaces that mate, seal, or align. (e.g., a shaft diameter, a bolt circle, a sealing face). Specify these explicitly.
• Non-Critical “Free” Dimensions: General outer contours and features where variation does not affect function. These can have looser, implicit tolerances.

Step 2: Apply Design for Additive Manufacturing (DfAM) Principles

• Design in Clearances: For snap-fits, press-fits, and moving assemblies, design in larger clearances than for injection molding. Start with 0.3-0.5mm clearance per side for FDM/SLS, and 0.2mm for SLA/SLM.
• Use Alignment Features: Instead of expecting two flat edges to mate perfectly, add dowel pins and holes, tongue-and-groove, or other self-aligning geometry.
• Avoid Critical Features on Supported Surfaces: If a hole must be precise, orient the part so the hole prints without supports.

Step 3: Communicate with Your Provider or Process

• Provide a Proper Drawing: Even for 3D printing, a 2D drawing with critical dimensions and tolerances called out is the universal language of intent. Use GD&T where appropriate (e.g., positional tolerance for a set of holes).
• Discuss and Understand the “As-Printed” State: Clarify if tolerances apply to the part directly off the printer, after support removal, or after specified finishing. This is a major source of misunderstanding.
• Request a First Article Inspection (FAI) Report: For critical parts, a reputable provider will supply dimensional inspection data (CMM or scanner) for the first part in a batch, proving they can hit your numbers.

How to write a spec that gets priced correctly

A drawing is a contract. If it leaves a number open, the shop fills that number in with its own default. The default is rarely the one you had in mind. Put these five lines on every RFQ and most of the back-and-forth disappears before it starts.

• File format and units, with the model exported in millimetres
• Process and material grade, not a trade name
• Critical features and the tolerance each one needs
• Surface finish and the faces it applies to
• Quantity for the pilot build and for the year

The point is not paperwork. Each line removes a decision the shop would otherwise make for you. A complete RFQ comes back priced. An incomplete one comes back guessed.

What actually drives the price

Buyers usually push on material because it is the line they can see. In practice the material is rarely the biggest number on a quote. These five lines decide most of it:

• build volume and orientation
• layer height and support structure
• material grade and colour
• post-processing and hand finishing
• quantity per build plate

Ask for the quote split by line. A supplier who cannot break it down is guessing, and you will pay for the guess. The line you can usually renegotiate is the tolerance band, not the raw material.

Problems you will meet, and the fix

Every process has a failure mode it prefers. The same four show up again and again on this kind of work, and none of them are mysteries once you know what to look for.

• Warping on long flat faces where the part lifts off the plate
• Layer lines showing through a cosmetic surface
• Support scars on a mating face nobody flagged
• Dimensional drift on thin walls after cooling

Catch the fault on the first part, not the hundredth. A first-article check costs an hour. A bad batch costs the whole order. Write the check into the purchase order so nobody has to remember it.

What to check on the first article

The first article is the cheapest inspection you will ever buy. Use it properly and the rest of the run becomes boring, which is exactly what you want.

• Every dimension on the drawing, with the measured value written beside it.
• The material certificate, matched to the heat or lot number on the parts.
• The finish against the sample you approved, under the same light.
• Fit against the mating part, assembled rather than measured.
• Packaging, because parts that arrive damaged were never in tolerance.

Check in this order: datums, then mating features, then cosmetic faces. If the datums are wrong, every number after them is noise. Keep the report on file. When a problem shows up in month six, that document is the only thing that tells you what changed.

How to compare two quotes fairly

Two numbers are not comparable unless the scope matches. Line them up before you decide, because a low number with three gaps is not a low number.

Same drawing, same material, same finish, same inspection. Anything else is a different job wearing the same part number. Then look at what each quote leaves out. Freight, packaging, certificates and tooling are the usual gaps.

Ask each supplier to state their assumptions in writing. The one who does is usually the one who has run the process before.

Scaling from one part to a batch

The first part teaches you the process. The hundredth teaches you the process cost.

Watch three things as volume grows: setup amortisation, tool wear, and how much hand work the finish still needs. If the unit price barely moves between ten and a hundred, the cost sits in material or in hand work, not in setup. Ask which, because only one of them is negotiable.

Lead time behaves the same way. A one-off is scheduled around other jobs. A repeat order is scheduled around your forecast. Give the shop a forecast and the lead time gets shorter, not longer.

How to read a supplier's quote

A quote is a short document with a long shadow. Four things in it decide whether the job runs smoothly, and only one of them is the price.

• The scope line. It should name the process, the material and the finish, not just the part.
• The tolerance basis. General tolerance or the drawing, and which standard it follows.
• The exclusions. Tooling, freight, certificates and inspection are the usual omissions.
• The validity date. Material prices move, and an open-ended quote is a promise nobody can keep.

Read the exclusions before you read the total. A quote that leaves out tooling is not cheaper. It is a quote for a different job.

What to send with the first order

The first order sets the pattern for every order after it. Send a complete package once and the shop files it, and the second order takes a fraction of the time.

• A STEP or native model plus a PDF drawing carrying the same revision letter.
• The material callout by standard number, not by trade name.
• The finish and the exact faces it applies to.
• Any test the part must pass: fit, seal, drop, salt spray, dielectric.
• A contact who can answer a question inside one working day.

Then write down what was agreed. An email that confirms the material, the finish and the date is worth more than a longer purchase order that leaves them open.

The Professional Prototyping Perspective: Tolerancing as a Collaboration

In our role as a bridge between design and production, we see tolerance management as a collaborative, iterative discovery process.

• Case Study – The Misaligned Manifold: A client designed a complex fluid manifold with 10 perpendicular port connections, specifying ±0.1mm positional tolerance for all ports. They requested SLS Nylon. We analyzed the model and advised that, due to inherent powder cake cooling stresses, ±0.25mm was a more realistic as-printed target for the unsupported vertical ports. The options were: 1) Accept ±0.25mm and redesign seals, 2) Machine the ports post-print (adding cost), or 3) Use a lower-shrinkage, more expensive material. We produced test coupons with the port features, measured them, and provided the data. The client opted for option 1, adjusting their seal design based on real data, avoiding a failed assembly.
• The “Good Enough” Philosophy: The goal is fitness for purpose, not arbitrary precision. For a bracket holding a wire loom, ±1mm may be perfectly functional. Spending engineering and manufacturing effort to achieve ±0.1mm on that bracket is wasted resource. Allocate your tolerance budget to the features that matter.

Mastering 3D printing tolerances is about embracing and managing inherent variation, not fighting it. By understanding the material, machine, geometric, and post-processing factors that consume your tolerance budget, you can make intelligent decisions at the design stage. The most successful users treat the first print not as a final part, but as a dimensional test coupon. Measure it, understand how your specific design reacts to the chosen process, and iterate. By partnering with knowledgeable providers and applying a disciplined DfAM approach, you can reliably bridge the gap between your perfect digital design and a perfectly functional physical part.

Our solution for Realistic Tolerances

YiguProto runs 3D printing in house, from a single prototype to a production batch. We quote the process, the material and the finish as separate lines, so you can see where the money goes instead of arguing with one number.

Send a STEP file or a drawing and you get a DFM report in one to two business days. It flags the features that add cost and the features that will not hold tolerance on the process you picked.

Every order ships with an inspection report on the features you call out. If a part misses, we make it again. Ready to price a part? Start with a free quote and DFM review, or read more about our 3D printing. You can also talk to an engineer before you commit to a drawing.

FAQ about Realistic Tolerances

The short version

Realistic Tolerances rewards the buyer who reads the data sheet and the drawing together. Pick the option for the property that matters, not for the price per kilo.

Tighten only what mates. Test a sample before the run. Then let the process do the rest.