Learning effects exist with regard to various behaviors, and especially work-related processes. This study measures the performance of a software testing project with time-varying effects using a software reliability growth model (SRGM), and discusses the changes in learning effects parameters with change-points in the model. We employ Chiu's  model to construct the time-varying learning effects and measure the performance of the software testing project using the data set in Huang and Hung . This paper also discusses the time-lag between error-detected and error-removed that exists with different learning concepts. The results indicate that error-removed requires more cognitive learning process time, and this information can be used to help project managers mastering the staff and the process of software testing, efficiently.