How many rejects do I need before the comparison means anything?

At least five expected rejects per cavity at the mold's overall rate, the NIST criterion for the normal approximation. Below that, keep sorting by cavity for more shots, or use Fisher's exact test.

Why compare with the other cavities instead of the mold average?

Because the mold average includes this cavity and is pulled toward it. The other cavities alone ran the same shots, resin and settings, so they are the fair baseline.

Should I block off the outlier cavity?

If it is a real outlier and its rejects cost more than the lost output, block it until the next stop. Blocking changes shot size and fill balance, so reset the process for the remaining cavities.

What usually makes one cavity worse than the others?

Local causes: blocked vents that give short shots or burn marks, a damaged parting line that flashes, a plugged cooling line, or gate damage.

Injection Molding calculator

Defect Rate by Cavity Calculator

Find one cavity's defect rate and whether it is worse than the rest of the mold or just sampling noise. You need the shot count, cavities running and rejects sorted by cavity number.

What this calculator does

Reject rate for one mold cavity, tested against the other cavities to show whether it is a real outlier.

Formula used

Cavity defect rate = cavity rejects ÷ n₁ × 100, where n₁ = parts molded per cavity
Other cavities' rate = other rejects ÷ n₂ × 100, where n₂ = n₁ × (cavities − 1)
Pooled rate p = all rejects ÷ (n₁ + n₂)
z = (cavity rate − other rate) ÷ 100 ÷ √(p × (1 − p) × (1 ÷ n₁ + 1 ÷ n₂)); outlier when z ≥ 1.645
Rejects above the other cavities' rate = cavity rejects − n₁ × other cavities' rate ÷ 100

Inputs explained

Rejects From This Cavity: Scrap sorted by the cavity number molded into each part.
Parts Molded per Cavity: Shots in the period; each cavity makes one part per shot.
Cavities Running: Cavities that filled every shot; leave out blocked cavities.
Rejects From All Other Cavities: Rejects from the remaining cavities over the same shots.

How to use the result

Best suited to deciding whether to block off a cavity, ranking cavities for the next mold PM.
Testing every cavity in turn at 95% flags about one good cavity in 20 falsely; confirm on a fresh run. It shows which cavity, not why; the defect type points to gate, vent, cooling or parting line.

Current U.S. benchmarks

The producer price index for plastic resins and materials stands at 280.569 (BLS, Aug 2026), up 6.6% from a year earlier. Quotes priced off last quarter's material cost miss this move.
The U.S. has 9,635 plastics product manufacturing establishments employing about 677,302 workers (Census County Business Patterns, 2023).

Common questions

How many rejects do I need before the comparison means anything? At least five expected rejects per cavity at the mold's overall rate, the NIST criterion for the normal approximation. Below that, keep sorting by cavity for more shots, or use Fisher's exact test.
Why compare with the other cavities instead of the mold average? Because the mold average includes this cavity and is pulled toward it. The other cavities alone ran the same shots, resin and settings, so they are the fair baseline.
Should I block off the outlier cavity? If it is a real outlier and its rejects cost more than the lost output, block it until the next stop. Blocking changes shot size and fill balance, so reset the process for the remaining cavities.
What usually makes one cavity worse than the others? Local causes: blocked vents that give short shots or burn marks, a damaged parting line that flashes, a plugged cooling line, or gate damage.

Last reviewed 2026-10-01.