AbstractThe paper addresses the problem of assessing the accuracy and utility of automatic quality tests used in high-speed production. A method is proposed for estimating the probabilities that the production process produces a product which is defective, that non-defective product will pass the test, and that a defective product fails the test. Data for printed circuit manufacture are used to demonstrate the method. Models of the effectiveness of various product testing procedures are investigated, the expected net profit is calculated and the probability of dispatching a defective product to a customer assessed.