Last updated: 2026-09-05 · 5 min read

For most injection molded parts, ±0.1 mm is the practical standard tolerance, ±0.05 mm is achievable on selected dimensions with the right resin and mold, and ±0.02 mm and tighter belongs to precision tooling with documented conditions. Anything beyond that is not a molding question but a redesign question. The number on your drawing is only achievable if four things line up: resin shrinkage behavior, part geometry, mold quality and process stability — and the cheapest of the four is choosing which dimensions actually need the tight number, before any steel is cut and before the quotation is priced.

Key facts

<>

  • Practical tiers: ±0.1 mm standard · ±0.05 mm selected dimensions · ±0.02 mm precision tooling · tighter means redesign or secondary operations.
  • Resin shrinkage sets the floor: amorphous resins hold tighter than semi-crystalline ones.
  • Tolerance cost is not linear — halving the tolerance can multiply machining, inspection and scrap cost several times.
  • Mark only functional dimensions tight; everything else falls to the mold’s standard tolerance, stated in writing.
  • Conformance must be defined with the tolerance: measurement method, sample plan, report format at T1.

    • What actually determines a molded dimension?

    A molded dimension is not cut to size — it is the result of a chain of physical events, and every link in the chain adds variation. The cavity is machined to a target; the resin fills it hot; then it shrinks as it cools, and keeps shrinking slightly after ejection. Mold temperature, melt temperature, packing pressure and cooling time each pull the final number in small ways. This is why injection molding tolerance is always expressed as a band, never as a single number, and why the same mold running the same resin will show a small spread across shots even in a well-controlled process.

    Plus/minus tolerance vs GD&T — A ± value bounds a single dimension. GD&T callouts (position, flatness, concentricity) bound relationships between features. Tight relationships often matter more than tight sizes for assembly — and they are what a dimensional report should actually measure against.

    Which tolerance bands are realistic?

    Tolerance band Realistic on Conditions required Typical application
    ±0.1 mm Most dimensions of most parts, up to moderate size. Sound mold, stable process, standard resin grade. Housings, brackets, consumer product internals.
    ±0.05 mm Selected critical dimensions under about 50 mm. Low-shrinkage resin, controlled mold temperature, tuned packing. Connectors, mating features, gear housings.
    ±0.02 mm Small, well-defined features on precision tools. Precision-machined hardened mold, stable resin lot, documented process window. Optical components, medical device features, micro-connectors.
    Tighter Rarely by molding alone. Usually means redesign, different process or secondary machining. Discuss in DFM before any steel is ordered.

    Two honest caveats. First, larger dimensions carry larger absolute variation — a ±0.05 mm band on a 10 mm feature and on a 100 mm feature are different problems. Second, these are production bands with a capable process, not one lucky sample at T1; a single in-spec shot proves nothing about the next hundred thousand.

    Why does resin shrinkage set the floor?

    Shrinkage is the physical fact behind every tolerance conversation, and resin families behave very differently:

  • Amorphous resins such as ABS, PC and PS shrink less and more predictably — they are the friendly territory for tight dimensions.
  • Semi-crystalline resins such as PP, PA and POM shrink more, and their shrinkage depends on cooling behavior, wall thickness and packing — tighter bands here demand more process control, sometimes different design targets.
  • Filled grades change the picture again: glass fiber cuts shrinkage dramatically but makes it directional — different along the flow direction than across it. A dimension that crosses flow lines may need its own tolerance discussion.

    • This is why a tolerance question without a resin answer is unanswerable. The factory should state the shrinkage value it designs the cavity against, for the exact grade you name — and you should see that value in the DFM, not discover its consequences at T1.

    Which dimensions deserve the tight number?

    The fastest way to inflate a mold price is marking every dimension tight. The disciplined approach runs in four steps:

    1. List what the part mates with. Assembly interfaces, seals, snap fits, fastener bosses — these are the candidates.
    2. Ask what failure looks like per dimension. If a dimension drifting 0.15 mm would still assemble and function, it does not need ±0.05.
    3. Separate size from relationship. Often the hole position matters at ±0.05 while the hole diameter is fine at ±0.1 — GD&T callouts let you say exactly that.
    4. Leave everything else at standard. The quotation should state the mold’s default tolerance for unmarked dimensions; then scope is explicit instead of assumed.

    A useful test from the quoting side: when every dimension is marked critical, the factory must either price the whole part as precision work or assume you do not mean it. Both outcomes are worse than a drawing with ten marked dimensions and a stated default.

    How does the mold itself enable tight tolerance?

    When a dimension genuinely needs ±0.05 or better, the tool has to be built to enable it. This is where tolerance stops being a drawing note and becomes construction scope:

  • Machining capability and steel stability. Precision-ground hardened inserts hold feature size; soft steel polished to hope does not. The steel story is in our steel comparison.
  • Temperature control. Uniform, stable mold temperature is the difference between a repeatable shrinkage value and a wandering one — cooling layout matters more on tight parts than anywhere else.
  • Gate design and packing access. The gate must let packing pressure work on the critical zone; a gate that freezes early hands control back to luck.
  • Measurement at acceptance. Tight tolerance without a measurement plan is unverifiable — the quote should include how the dimension will be proven, with what instrument, on how many shots.

    • What does tolerance actually cost?

    Tolerance cost is not linear, and this is the part of the conversation most drawings ignore. Moving a dimension from ±0.1 to ±0.05 mm does not add a fixed percentage — it changes the construction (hardened inserts, better temperature control), the machining time (slower passes, more EDM), the trial loop (more tuning shots, more measurements) and the scrap risk during production. Each of those multiplies, and the total can run several times the difference people imagine. The engineering response is not to avoid tight tolerance — it is to spend it precisely where function demands it, and nowhere else.

    Capability, not heroics — a tolerance is affordable when the process can hold it routinely. If a dimension only hits spec during a lucky window at T1, the tool will fight the production team for years. Ask for evidence of stable holding — consecutive shots, not best shots.

    What happens when tolerance is skipped at T1?

    Tolerance questions that are not settled at T1 do not disappear — they migrate downstream, where each move makes them more expensive:

    1. At T1, nobody measures. Samples “look right”, the tool ships, and the dimensional question becomes an untested assumption carried across an ocean.
    2. At first assembly, parts fight. Snap fits resist, seals leak, fasteners bind. The production line starts sorting good parts from bad ones — labor replacing tooling, at the worst possible rate.
    3. The diagnosis crosses borders. Is it the mold, the resin lot, or the process at your molder? Without a baseline report, the argument starts from zero, and every week of it costs more than the measurement would have.
    4. The fix arrives as a change order. Mold modification at this stage is slower and pricier than at T1 — the tool may need inserts reworked or replaced entirely, with shipping cycles in between, and the leverage to negotiate is mostly spent.

    The pattern is consistent: every tolerance verified at T1 with a signed report costs a few measurement hours; every tolerance discovered in production costs weeks and a renegotiation. This is the same evidence-first principle as the rest of mold sourcing — the report is not paperwork, it is the cheapest insurance available in the project.

    How should tolerances be written into the RFQ?

    Tolerance is one of the nine blocks of a complete RFQ, and the same rule applies: nothing left to assumption. Concretely: mark the critical dimensions on the 2D drawing; state the band or GD&T callout for each; name the resin grade so shrinkage is calculable; define the measurement method and sample plan; and require the factory to state the default tolerance for everything unmarked. With that in place, the dimensional report at T1 becomes a real acceptance document — measured values against named callouts — instead of a courtesy attachment. The full document structure is in the RFQ checklist.

    The drawings that cost the most are the ones where every dimension is marked critical, because they tell me the buyer has not decided what actually matters. Give me ten marked dimensions and a stated default for the rest, and I can price honestly, build the right inserts and prove the numbers at T1 with a report you can argue with. Tolerance is a budget — spend it where your assembly needs it, and keep the receipt in the measurement plan.

    — Arthur, Tooling Engineer at aCatMould

    Frequently asked questions

    What is a standard injection molding tolerance?

    For general dimensions on common resins, ±0.1 mm is the practical working standard. Anything tighter should be a deliberate, dimension-by-dimension decision with the cost understood, not a default copied from a machining drawing. State the default tolerance for unmarked dimensions in the RFQ so both sides price the same scope.

    Can molding hold the same tolerances as CNC machining?

    Not across the part, and not at the same cost. Molding reaches tight numbers on selected small features with precision tooling; machining holds them everywhere but pays per part. For features that truly need machining-level numbers, the honest options are redesign, secondary operations or a different process.

    Does part size change achievable tolerance?

    Yes. Shrinkage is proportional, so a long dimension accumulates more absolute variation than a short one. Tolerance bands on drawings should scale with feature size — a flat ±0.05 across a 200 mm part is rarely realistic.

    Which resins hold tolerance best?

    Amorphous resins like ABS and PC shrink less and more predictably, and glass-filled grades cut shrinkage further — at the price of directional behavior. Semi-crystalline families like PP and PA are more demanding. Always tie the tolerance discussion to the specific grade.

    How is tolerance conformance proven at T1?

    With a dimensional report against your drawing: measured values per callout, on consecutive shots from a stable process, with the instrument and method named. One in-spec sample is evidence of possibility; a multi-shot spread is evidence of capability.

    Can tight tolerances be added after the mold is built?

    Sometimes, at a price. Steel-safe design helps — features can often be tightened by machining inserts smaller, but opening a dimension is limited by how much stock exists. This is exactly the kind of risk a DFM review should flag before the deposit, not after.

    Key takeaways

    1. Realistic tiers: ±0.1 mm standard, ±0.05 mm on selected dimensions, ±0.02 mm with precision tooling — tighter usually means redesign.
    2. Resin shrinkage sets the floor: amorphous resins and filled grades behave differently, and the tolerance question is unanswerable without the grade.
    3. Mark only functional dimensions tight and state a default for the rest — an all-critical drawing prices as precision everywhere or gets ignored.
    4. Tight tolerance becomes construction scope: hardened precision inserts, stable temperature control, gate and packing design, and a measurement plan.
    5. Accept on capability evidence — consecutive-shot measurements at T1 against named callouts — not on a single lucky sample.

    If your drawing carries tolerances you want verified rather than hoped for, send us the part data and the marked dimensions. You will get a written DFM with the shrinkage values we would design to, a line-item quotation and a T1 measurement plan — from a named engineer.

    Sources
    ISO 20457:2018 — Plastics moulded parts: Tolerances and acceptance conditions
    ASTM D638 — Standard Test Method for Tensile Properties of Plastics, ASTM International
    MatWeb — material property database for resin and steel datasheets
    Plastics Industry Association — SPI mold standards and surface finish sets

    Sources: Plastics Technology · Plastics Technology Online · Society of Plastics Engineers (SPE) · ThomasNet