Typical mold shrinkage runs about 0.3 to 0.8 percent for amorphous plastics such as ABS and polycarbonate, and 1 to 4 percent for semi-crystalline ones such as polypropylene, polyethylene and acetal. Glass-filled grades shrink less along the flow than across it. Toolmakers enlarge the cavity per direction, mold a trial, measure, and then correct the steel.
Plastic goes into the mold hot and compressed and comes out cooler, denser and smaller. The amount is reported as mold shrinkage: the difference between the cavity size and the part size, divided by the cavity size, measured on a standard specimen. ASTM D955 and ISO 294-4 are the two test methods behind most datasheet values. Two details in those methods change how you should use the number. ASTM D955 measures at 24 and 48 hours after molding, because parts keep moving after they leave the press. And the standard says plainly that its results give comparable data on standard specimens but cannot predict absolute values in real parts with different flow paths, walls, pressures and temperatures.
So a datasheet shrink value is a good comparison between grades and a starting point for the toolmaker. It is not a promise about your part. BASF's datasheet for Ultramid A3WG6 shows the gap clearly: the ISO 294-4 plaque gives 0.38% along the flow and 0.93% across it, while a constrained test box molded under stated conditions (melt 290 °C, mold 80 °C, 1.5 mm wall) gives 0.49%. Three numbers, one grade.
The biggest divider is structure. Amorphous resins (ABS, PC, acrylic, polysulfone) have no crystals to form as they cool, so they shrink less and roughly the same in every direction. Semi-crystalline resins (PP, PE, nylon, acetal, PBT) pack their chains into crystals on cooling, so they shrink more, and the amount depends on how fast they cool and how the chains were oriented by the flow. RTP Company's part design guide gives these typical values for a 3.2 mm (1/8 in) section tested to ASTM D955:
| Resin | Family | Typical shrink, in/in | Same as % |
|---|---|---|---|
| ABS, unfilled | Amorphous | 0.004 - 0.008 | 0.4 - 0.8 |
| Polycarbonate, unfilled | Amorphous | 0.003 - 0.007 | 0.3 - 0.7 |
| Acrylic (PMMA) | Amorphous | 0.003 - 0.007 | 0.3 - 0.7 |
| Polypropylene, unfilled | Semi-crystalline | 0.010 - 0.025 | 1.0 - 2.5 |
| Polypropylene, talc filled | Semi-crystalline | 0.007 - 0.015 | 0.7 - 1.5 |
| HDPE | Semi-crystalline | 0.015 - 0.040 | 1.5 - 4.0 |
| Nylon 6/6, unfilled | Semi-crystalline | 0.008 - 0.015 | 0.8 - 1.5 |
| Acetal (POM) | Semi-crystalline | 0.020 - 0.025 | 2.0 - 2.5 |
Those are ranges across many grades; a single grade's datasheet is narrower and is the number the toolmaker should use. The site's plastic shrinkage calculator carries a fuller table and does the cavity arithmetic for one dimension at a time.
Fillers and oriented chains make shrinkage directional. Glass fibers line up with the flow and refuse to shrink along their length, so a filled part shrinks much less along the flow than across it. RTP's guide calls this anisotropic shrinkage and notes that filled materials usually show it. An unfilled amorphous grade barely does: Covestro lists Makrolon 2405 polycarbonate at 0.65% parallel and 0.7% normal. At the far end, Celanese's Vectra LCP guide says shrinkage in the flow direction can be virtually zero, because the rigid molecules orient so strongly.
A hypothetical tray in 30% glass PA66 measures 120 mm long by 60 mm wide, gated at one end so the melt runs along the 120 mm length. Using the BASF plaque values, 0.38% along the flow and 0.93% across it, the cavity is sized with the standard relation cavity = part / (1 - shrink):
Now suppose someone averaged the two values and scaled the model uniformly at 0.655%. The cavity would be 120.791 by 60.396 mm. Molded, the length would come out about 120.33 mm (0.33 mm long) and the width about 59.83 mm (0.17 mm short). On a part with a ±0.1 mm tolerance, both are rejects, and they miss in opposite directions, so no single process change fixes both. The same tray in the Covestro polycarbonate would need 120.785 by 60.423 mm, nearly the same factor on both axes.
The lesson is not that 0.38 and 0.93 are the right numbers for your tray. Flow direction in a real part curves, splits and turns at ribs, and the plaque values are a starting point. The lesson is that a filled part needs a shrink value per direction, chosen from a fill pattern, not one number typed into a scale command.

Holding pressure and time pack extra material in and lower shrink; mold temperature and wall thickness change cooling rate and, for semi-crystalline grades, crystallinity. RTP's guide flags thick sections and non-uniform walls as the source of shrink differences that lead to warp. That is why shrink and warp are the same problem seen from two angles; why parts warp covers the second one, and wall thickness covers the design rule that keeps both under control.
A large flat part in a semi-crystalline resin is the hardest case: high, directional shrink plus warp. A small amorphous part with loose tolerances is the easiest; one shrink factor and a trial usually do it. LCP is its own world, with near-zero flow shrink, so the Celanese guide's advice to work from the flow pattern applies more strongly than ever.
The value depends on the test specimen, wall thickness, molding conditions and whether it was measured along or across the flow. ASTM D955 and ISO 294-4 use different specimens, and suppliers report conditions in footnotes. Compare grades only on the same test basis, and treat any datasheet value as a starting point for the toolmaker.
Less overall, but unevenly. The fibers line up with the flow and resist shrinking along their length, so shrink along the flow can be less than half the shrink across it. That directional difference, not the total, is what causes most dimensional and warp trouble with filled parts.
A little. More holding pressure and time, or a different mold temperature, can move a dimension by a small amount, but pushing the process to its edge to hit size leaves no room for normal variation. The usual practice is to set a stable process first and then correct the steel to match it.
Yes, mostly in the first day or two, which is why ASTM D955 measures at 24 and 48 hours. Semi-crystalline resins move more than amorphous ones. Nylon also absorbs moisture and changes size afterward, so its inspection plan should state whether parts are measured dry or conditioned.
If the new grade's shrink differs, the parts come out a different size from the same steel. Switching from ABS to polypropylene, for example, multiplies the shrink by two to three times. Even a change between similar grades, or a different glass content, should be checked with a trial and a full dimensional report before production.
Send the part file and the quantity you expect to run, and we'll review it before you commit to tooling. NDA first if you want one.
Send a part for reviewBook a 30-minute call