Impact of reliability growth on repairable inventory with the application to semiconductor test equipment is discussed and analyzed in this paper. Semiconductor test equipment is capital-intensive product built with swappable and repairable modules (i.e. printed-circuit-boards) to facilitate the maintenance and repair of the system. A stochastic inventory model is proposed to estimate the demand rate for defective modules considering the reliability growth and the increase of field system installations. Three test statistics, the Laplace test, the nonparametric test and the Crow/AMSAA, are used to examine the reliability growth trend for repairable systems: homogeneous Poisson failures (HPP) vs. non-homogenous Poisson failures (NHPP). For NHPP, a modified Crow estimate is used to generate the estimate of the module failure intensity, based on which the demand for spare modules is forecasted by considering the installation rate of new systems in the field. To obtain the repair rate estimate for defective modules, the uncertainty in the transition time of the defective modules to the repair center and the product failure modes are appropriately combined to estimate the mean and the variance of the repair rate. Then the service quality index is derived based on the repair rate and the expected failed modules from field systems. Finally a computerized inventory management system is automated using PC-based MS Visual Basic and Excel programs.