Neat ABS shrinkage is documented at 0.4–0.7% (density 1.04–1.08 g/cm3) and rises to 0.7–1.6% in the broadest published ABS range, while neat PBT sits at 1.2–2.5% (density 1.30–1.43 g/cm3) per published polymer tables [S1][S4].
20% glass-filled ABS drops to 0.1–0.2% and 30% GF ABS to 0.2–0.3%; 20% glass-filled PBT/ABS blend is reported at 0.3–0.5%, with 20% GF PBT in a similar low-shrink window [S1][S8]. The difference tracks amorphous versus semi-crystalline structure, not barrel temperature or hold pressure, so material selection decides most of the dimensional outcome before the first shot is fired.
Why the ABS versus PBT gap exists in the first place
ABS is amorphous, so its melt contracts only on thermal cooling; published figures place neat ABS at 0.4–0.7% with FR grades dipping to 0.3–0.6% and high-heat grades at 0.4–0.9% [S1][S7]. PBT is semi-crystalline, and crystallite packing pulls the macromolecular chains into a denser arrangement, lifting neat PBT shrinkage to 1.2–2.5% and driving UL-HB flame-retardant PBT into the same 1.2–2.5% band [S2][S4]. The mechanism is the same one that separates PP (1.0–2.5%) and HDPE (1.5–3.0%) from PC (0.4–0.6%) in any side-by-side shrinkage table [S2].
The practical read is that any 100 mm PBT dimension can lose 1.2–2.5 mm off the cavity, while the same dimension in neat ABS loses 0.4–0.7 mm, and glass reinforcement tightens that further: 20% GF ABS lands at 0.1–0.2% and 30% GF ABS at 0.2–0.3% [S1][S4]. When the part is dimensionally critical, the choice between amorphous and semi-crystalline is the single largest lever, and hold pressure, melt temperature, and cooling time fine-tune from there.
Side-by-side shrinkage numbers for the grades a buyer actually sees
The cleanest comparison is from the same published table so the test conditions (typically ASTM D955 or ISO 294-4) are consistent [S1][S2][S4]:
Neat ABS: 0.4–0.7% (density 1.04–1.08 g/cm3). ABS flame-retardant: 0.3–0.6%. ABS high-impact and high-heat: 0.4–0.9% each. ABS/PC blend: 0.5–0.7%. ABS/PC 20% GF: 0.2–0.3%. ABS (20% GF): 0.1–0.2%. ABS (30% GF): 0.2–0.3% [S1][S4].
Neat PBT: 1.2–2.5% (density 1.30–1.43 g/cm3). PBT/ABS blend: 0.5–0.7% (density ~1.30 g/cm3). 20% glass-fiber-reinforced PBT/ABS: 0.3–0.5%. PC/PBT blend: 0.5–0.7% (density 1.20–1.26 g/cm3) [S1][S4][S8].
Cross-checking the engineering rules: in the amorphous column, ABS, ABS/PC, PC/ABS, and ASA all cluster between 0.4% and 0.7% neat, with the glass-filled variants from any of those families compressing toward 0.1–0.3% [S1][S4]. In the semi-crystalline column, neat PBT, PA6 (1.0–1.8% MD / 1.3–2.2% TD), and POM homopolymer (1.8–2.5%) all sit in the high-shrink window, and 20–30% glass reinforcement cuts them by roughly a factor of three to five [S4]. That cross-check is what tells you the PBT number is the real number, not a test artefact.
Decision matrix: pick by tolerance, environment, and finish
For tight-tolerance cosmetic parts under 0.3% shrinkage, specify ABS or ABS/PC neat, or 20% glass-filled PBT/ABS at 0.3–0.5% when higher heat or chemical resistance is needed [S1][S4][S8]. For automotive interior trim, appliance housings, and consumer electronics, neat ABS at 0.4–0.7% is the default low-risk choice and pairs cleanly with existing molding line flow paths.
For under-hood, connector, and electrical parts where HDT and chemical resistance dominate, 20% GF PBT or PC/PBT at 0.5–0.7% is the workhorse, accepting the higher base shrinkage in exchange for stiffness and creep performance [S4]. For high-flow thin-wall parts on an automatic molding line, ABS/PC 20% GF (0.2–0.3%) gives both the flow length and the dimensional repeatability needed to hold ±0.05 mm on a 50 mm feature [S1][S4].
Avoid neat PBT when the drawing calls for ±0.1 mm on a feature larger than 20 mm: a 1.2–2.5% shrink range means the same cavity can ship at 0.5% or 2.0% depending on hold pressure, and that spread is wider than the tolerance band in most cases [S2][S4].
Process levers that move the needle after material is fixed
Holding pressure is the strongest secondary control: higher and longer hold pressure packs more melt into the cavity as the skin freezes, lowering the effective shrinkage for both ABS and PBT, but the absolute shrinkage of PBT still stays above filled ABS in every published datapoint [S3]. Cavity wall temperature moves the moulded-skin crystallinity in PBT, so a 10–20°C bump on the cavity side can shift PBT shrinkage by 0.2–0.5 percentage points, which is material-side chemistry, not just thermal expansion [S3].
Wall-thickness variation is the third lever and the one most often missed: a 2 mm section in ABS cools with 0.4% shrink, while a 4 mm section in the same mould runs closer to 0.7%, and the same scaling applies to PBT in the higher band [S3][S5]. Reinforcing ribs and glass fillers act as mechanical constraints, lowering local shrinkage to 0.1–0.3% in filled grades and pulling neat-polymer zones toward 0.5–0.7% [S3][S4]. For shop-floor quoting, the rule of thumb is to design ribs at 60% of nominal wall and verify on a test coupon per ASTM D955 before cutting production steel [S2].
Standards and test methods that anchor the numbers
Moulding shrinkage is measured against the cold cavity dimension, and the two reference methods are ASTM D955 (the US default) and ISO 294-4 (the international default for thermoplastics), with ISO 2577 covering thermosets [S2]. Both report shrinkage as a percentage of the cavity dimension, and both require a conditioned test specimen, so any number quoted without a method tag is a marketing number.
For datasheet work, density 1.04–1.08 g/cm3 for ABS and 1.30–1.43 g/cm3 for PBT is the published bracket, and that density delta is why PBT sinks in a water-filled QA tank while ABS floats, a quick on-floor check that often catches material mix-ups before the dimensional inspection does [S4]. The 0.5–0.7% band for PBT/ABS and PC/PBT blends reflects the amorphous ABS phase partially suppressing PBT crystallite growth, which is also why PC/PBT at 1.20–1.26 g/cm3 sits between neat PBT and neat PC on the table [S4].
Use cases and where the two polymers misbehave
ABS is the safe pick for indoor consumer-goods housings, LEGO-style snap-fit assemblies, and 3D-printed jigs, because the 0.4–0.7% neat band keeps warp and sink under control even with non-uniform wall sections [S1][S5]. It is the wrong pick for sustained service above 100°C, continuous UV exposure without a cap layer, or any application calling for high chemical resistance to fuels and oils, where PBT or glass-filled PBT takes over despite the higher shrinkage penalty [S2][S4].
PBT in its neat form is the wrong pick for cosmetic parts with ±0.05 mm tolerances, and the shrinkage can drift by 0.3–0.8% between shots if hold pressure and cooling vary, which is the most common source of dimension rejections on a shell molding machine running PBT for the first time [S3]. Glass-filled PBT and PBT/ABS blends close that gap to 0.3–0.7% and are the realistic choice when the part must survive under-hood temperatures and still hold a press-fit dimension [S1][S4][S8]. For high-pressure die-casting-like shot consistency on a static-pressure molding machine, 20% GF PBT/ABS at 0.3–0.5% is the most forgiving combination in the published tables [S1][S8].
Cost, lead time, and shop-floor reality
Neat ABS pellets are commodity-priced and ship in 1–3 weeks from regional distributors, while neat PBT and glass-filled PBT/ABS blends carry a 20–40% resin premium and longer lead times driven by glass-fiber supply and the compounding step [S4][S5]. The compounding step is also why lot-to-lot shrinkage in filled PBT/ABS is tighter than in neat PBT: a 0.3–0.5% published range beats a 1.2–2.5% range for SPC charts, even if the per-pound cost is higher [S1][S4][S8].
For new tooling on a molding cell, the realistic recommendation is to draft the cavity within the published shrinkage band for the target material (for example, 0.4-0.7% for ABS and 0.3-0.5% for 20% glass fiber reinforced PBT/ABS), cut a test coupon per ASTM D955 or ISO 294-4, measure the actual shrinkage on the production machine, and adjust the cavity before the second mould tryout. That sequence keeps the steel cut to a single iteration in most cases and avoids the rework penalty that comes with guessing the high end of the PBT band [S3][S5].
Two trackable signals for the next design review: shrinkage tables on supplier datasheets moving toward tighter bands for 30% GF PBT/ABS (currently reported at 0.2–0.4% in industry tables), and mould-flow software updates shipping updated pvT datasets that predict PBT crystallinity at the cavity wall, both of which would compress the published ±0.5% PBT uncertainty band further [S3][S4].
For related coverage, see Aluminum Ladder Sizing for a 12 ft Ceiling: 8 ft vs 10 ft vs 12 ft Step Ladder.