This paper considers the development of a cost based industrial quality control system which functions both as an acceptance sampling plan and as a means for determining an assignable cause of quality variation in the manufacturing process. The manufacturing faci1ity studied normally produces items at a relatively constant proportion defective. The facility is subject, however, to random failures in time which cause the proportion defective to shift upwards. The objective of the paper is to develop a sampling plan which minimizes the expected total annual cost of quality control by determining t)he sample size, acceptance number, and time period between successive samples. The optimization technique employed is a combination of a multivariate search technique and a partial enumeration.