Optimal Design of A Degradation Test Sheng-Tsaing Tseng Institute of Statistics National Tsing-Hua University Hsin-Chu, Taiwan, ROC Email:sttseng@stat.nthu.edu.tw Key words- Highly-reliable products; Inspection frequency; Termination time; Stochastic diffusion process. Introduction Due to the market competition, how to provide the customers with the product’s lifetime information is a great challenge to the manufacturer. However, for highly reliable products, it is difficult to assess the lifetime of the products by using traditional life tests that record only time-to-failure. Even using the technique of censoring and/or accelerating the life by testing at higher levels of stress such as elevated temperatures or voltages provide little help, because no failures are likely to occur in a reasonable amount of time. If there exist quality characteristics whose degradation over time can be related to reliability, then collecting “degradation data” can provide information about product reliability. Nelson (1990, chapter 11) surveyed literature on this subject. Lu & Meeker (1993) proposed a method which used a degradation model to estimate the failure time distribution of a product. Carey & Koenig (1991) described a data analysis strategy and a two-stage model-fitting method to extract reliability information from observations on the degradation of integrated logic devices that are components in a new generation of submarine cables. Tseng et al. (1995) used a degradation model to improve the reliability of the fluorescent lamp. Tseng & Yu (1997) and Yu & Tseng (1998) proposed a method to determine the termination time for a degradation test and an ADT, respectively. In conducting a degradation experiment, the restriction of experimental cost and the precision of estimating the lifetime information of the product are two major concerns of the experimenter. Several decision variables are related to these two subjects. Among them, the sample size, the inspection frequency, and the termination