How Much Does Injection Molding Shrink?

Last updated: 2026-09-01 · Reading time: about 10 min

Almost every injection-molded part gets smaller as it cools — typically by 0.3 to 3 percent — and the mold is cut oversize on purpose to compensate.
Amorphous resins such as ABS, PC and PMMA cluster near the low end (roughly 0.3–0.8%), while semi-crystalline resins such as PP, PA and POM sit at the high end (roughly 1–3%). The exact number comes from the grade datasheet, and the part you actually measure depends on wall thickness, packing pressure, mold temperature and geometry — which is why shrinkage is a tooling decision made at DFM, not a correction applied after T1.

Key facts

<>

  • Molding shrinkage is the difference between the tool dimension and the part dimension after cooling — expressed as a percentage, and compensated by cutting the tool larger.
  • Amorphous resins shrink less and more predictably; semi-crystalline resins shrink more because crystallites form as the melt freezes and pull the material together.
  • Shrinkage is not one number: it differs with flow direction, wall thickness and packing, and glass fibers push the difference further apart.
  • Polyamides add post-molding shrinkage — moisture absorption and conditioning move dimensions for days after molding, on top of the thermal shrink.
  • The datasheet gives the range, but the process and the tool decide where in that range the part lands — which is why tolerance strategy and shrinkage strategy are one conversation.

    • Gate vestige — The visible mark left on the part surface where the gate was trimmed. Edge gate: 0.5–2.0mm vestige. Pin-point gate: 0.2–0.5mm. Valve gate: 0.05–0.1mm (flush). Vestige size affects cosmetic appearance and may require secondary trimming. See also related guide.

    What is molding shrinkage — and what is it not?

    Molding shrinkage
    The dimensional difference between the mold cavity and the cooled part, expressed as a percentage of the mold dimension. Tools are cut oversize by this factor so the finished part measures to the drawing.

    The definition sounds mechanical, and the first confusion always follows from taking it that way. Molding shrinkage is not a material constant like density — it is a behavior with a range, and the range is wide enough that two tools running the same resin in the same plant can produce parts that differ measurably.
    What the datasheet supplies is the envelope the grade has been tested to; what the tool supplies is the geometry, the gate position and the cooling layout; what the process supplies is packing pressure and time, which decide how much material is pushed into the cavity before the gate freezes.
    Shrinkage is also not reversible after the fact: once the steel is cut, the compensation is baked in, and pulling a part back to the drawing by reworking the process has limits that reinforced resins quickly. That is why a serious DFM conversation settles the shrinkage assumption before a cutting tool touches steel, not after the first shots measure off.

    What are typical shrinkage rates by resin family?

    The table below collects the working ranges molders use for orientation. Read them as family-level orientation only — the datasheet of the exact grade governs, and within one family the spread between grades can be as wide as the spread between families.

    Resin family Typical shrinkage range Why it sits there
    ABS ≈ 0.4–0.7% Amorphous; no crystallization pull, low and predictable.
    PC ≈ 0.5–0.7% Amorphous; stiff chains resist densification.
    PC/ABS ≈ 0.5–0.7% Blend tracks its amorphous parents.
    PMMA ≈ 0.3–0.8% Amorphous; optical grades vary with molecular weight.
    PP ≈ 1.2–2.5% Semi-crystalline; high crystallinity, strong packing sensitivity.
    PA6 / PA66 ≈ 1.0–2.0% Semi-crystalline; plus post-molding moisture movement.
    PA + GF30 ≈ 0.2–0.6% (directional) Fibers lock the flow direction; anisotropy replaces magnitude as the problem.
    POM ≈ 1.8–2.5% Highly crystalline; among the largest shrinkages in common use.
    PBT ≈ 1.2–2.0% Semi-crystalline; fast-freezing grades trend lower.

    Two habits in this table deserve attention. First, the glass-filled line: filling PA or PBT with fibers cuts the overall shrinkage but introduces anisotropy — the part shrinks less along the flow direction and more across it, so a flat panel dishes and a long boss walks off-center.
    Second, POM: gears and precision mechanisms specify it for its friction and stiffness, then have to live with shrinkage near 2%, which is exactly why gear tolerances and gear steels are discussed together in a POM versus nylon decision rather than separately.

    Why does the same resin shrink differently on different tools?

    Because shrinkage is the end of a chain that starts in the cavity. Wall thickness comes first: thick sections freeze slowly, hold more crystallinity in semi-crystalline resins, and shrink more than thin walls of the same resin — which is why ribs and bosses sink while the panel around them holds.
    Packing pressure comes second: until the gate freezes, holding pressure keeps feeding material to replace what shrinking pulls away, so a generous packing profile measurably lowers the final shrinkage, and a short one lets it run to the top of the datasheet range. Mold temperature plays the same card for crystalline resins: a hotter mold freezes slower, crystallizes more, and shrinks more.
    Finally, geometry and gating create direction — material oriented by flow shrinks differently along and across that flow, and the effect is modest in amorphous resins and central in filled ones. None of these levers is exotic; together they explain why the same PA66 datasheet covers a range rather than a point, and why the tool and process choices determine where inside that range the parts land.

    How do you shrink-compensate a mold?

    1Take the datasheet range, not a single number. Enter the low and high values for the exact grade into the DFM record — the range is the honest input.
    2Split critical features from cosmetic surfaces. Tight-tolerance features get the middle of the range with a sensitivity note; free tolerances take the family value without ceremony.
    3Account for direction on filled grades. For glass-filled resins, apply different factors along and across flow — and place gates with that anisotropy in mind, not after it.
    4Cut the steel at the compensated size. Cavity and core dimensions are scaled by the chosen factors, and the factors are recorded on the tool drawing.
    5Measure at T1 and reconcile. The full-dimensional report compares parts to drawing; where a feature measures off, the reconciliation decides whether process packing or steel moves — in that order of cost.

    Which geometries turn shrinkage into a problem?

  • Thick-to-thin transitions — a heavy boss feeding a thin wall shrinks at two rates in one part; uniform walls are the first defense.
  • Long flat panels — differential shrinkage across the panel shows up as warp, which is why polyamide flat parts and conditioning are discussed together.
  • Precision hole positions — shrinkage moves hole centers toward the mass; matched pairs drift apart in semi-crystalline resins unless compensated directionally.
  • Gear and mechanism features — POM-class shrinkage near 2% plus anisotropy needs the compensation split by feature, not one global factor.
  • Parts that gate from one edge — the flow-oriented axis and the cross axis shrink differently; flag it whenever a filled resin fills a flat part from one side.
  • Anything tolerance-critical near a gate — local overpacking distorts the local shrinkage; the datasheet cannot see your gate.

    • What is post-molding shrinkage?

    Some shrinkage keeps running after the part leaves the machine, and pretending otherwise is how programs discover their dimensions have drifted a week after approval.
    Two mechanisms drive it. Thermal relaxation: parts molded hot and ejects that cool unevenly continue to move as they room temperature — measured dimensions stabilize within hours to a day, which is why measurement timing is standardized. Moisture uptake: polyamides absorb water from the air after molding, and the swelling and conditioning movement runs over days — dimensions measured dry on a PA part will not be the dimensions it carries in service.
    The practical consequences are procedural, not mysterious: measure at the standardized time after molding, store PA parts in the state the drawing assumes (dry as molded, or conditioned), and treat any dimensional argument on polyamide with the moisture history on the table. Post-molding shrinkage does not defeat tolerancing — ignoring it does.

    What does the shrinkage conversation look like at DFM?

    In a disciplined program the conversation is four sentences long. The grade datasheet sets the range. The DFM review assigns the compensation per feature class — middle of range for tight features, family value elsewhere, directional factors for filled resins. The tool drawing records the factors so the assumption is inspectable, not folklore. And the T1 report measures the result, closing the loop between assumption and reality before the tool ships.
    That loop is also where dimensional standards earn their keep: tolerance tables for molded parts, such as ISO 20457, grade what is realistic per material group and feature, so the drawing asks for what the resin and geometry can deliver.
    When a program skips the loop — cuts steel on one number, discovers off-drawing parts at T1, and starts the process-versus-steel argument — the cost is not the rework itself; it is the calendar the rework consumes, and the credibility it spends with everyone who signed the drawing.

    “Shrinkage is the one assumption that hides inside every dimension on the drawing. We would rather argue about it on paper at DFM, where changing your mind is free.”
    — Arthur, Tooling Engineer at aCatMould

    Frequently asked questions

    Why not just cut the tool to the drawing and fix it at T1?

    Because the fix at T1 is steel or nothing. If the part measures off, your options are process packing (limited by the gate and the range), adding steel where it can be welded, or cutting cavities again. Compensating at DFM costs a spreadsheet row; discovering it at T1 costs weeks.

    Do I use the same shrinkage factor for every feature?

    Not for precision parts. Tight features get the range midpoint with a sensitivity note; filled resins get directional factors; free tolerances take the family value. One global factor is fine for loose tolerances and wrong for anything that measures.

    Does higher packing pressure fix shrinkage?

    It lowers shrinkage within limits — until the gate freezes, packing feeds material, so more packing time and pressure pull the part closer to the cavity size. But packing cannot beat a wrong compensation in thick sections, and excessive pressure buys flash and residual stress instead.

    How long after molding should I measure dimensions?

    Follow the standard your program uses — most specify measurement at a fixed interval after ejection, once thermal relaxation has settled. For polyamides, also record whether parts were measured dry or conditioned, because the two states differ measurably.

    Does glass fiber reduce shrinkage?

    It reduces the average and spreads the difference: along the flow direction shrinkage drops sharply, across it less so. The result is anisotropy — warp and dishing in flat parts — so filled grades need directional compensation and gate placement decided at DFM.

    Where is the shrinkage value recorded?

    On the tool drawing and in the DFM record, per feature class where it matters. If the factor lives only in someone’s head, the next modification or tool transfer starts from a guess — and the T1 report will quietly prove it.

    Key takeaways

    1. Shrinkage runs roughly 0.3–0.8% for amorphous resins and 1–3% for semi-crystalline ones — the grade datasheet sets the envelope.
    2. Where the part lands inside the range is decided by wall thickness, packing, mold temperature and direction — tool and process choices, not fate.
    3. Compensate at DFM per feature class, record the factors on the tool drawing, and reconcile against the T1 full-dimensional report.
    4. Filled resins trade magnitude for anisotropy — direction and gating become part of the shrinkage decision.
    5. Polyamides keep moving after molding; measure at the standardized time and state, or the numbers argue back.

    If your drawing carries tolerances that matter — and it does — send us the 3D model, the resin grade and the tolerance stack before the deposit. You will get the shrinkage compensation plan per feature class in writing, the directional strategy for any filled grade, and the T1 evidence list that reconciles assumption against measurement — from a named tooling engineer, before any steel is ordered.

    Sources
    ISO 20457:2018 — Plastics moulded parts: Tolerances and acceptance conditions, International Organization for Standardization
    MatWeb — material property database for grade-level shrinkage data
    Plastics Industry Association — industry resources on plastics processing practice
    ASTM D638 — Standard Test Method for Tensile Properties of Plastics, ASTM International

    Sources: Plastics Technology · Plastics Technology Online · Prospector — Resin Datasheets · MatWeb — Material Property Database