Data from the Scholastic Aptitude Test-Verbal (SAT-V), SAT Mathematics (SAT-M), and Achievement Tests in Biology, American History, and Social Studies were used for this study. The temporal stability of item parameter estimates obtained for the same set of items calibrated for different examinees at different times was analyzed. It was believed that greater time lapses in test administrations would result in greater differences between item parameter estimates obtained from test administration data. The type of test probably influences the stability of item parameter estimates. Parameter stability is affected by the fit of the data to the model. Aptitude test items were a better fit to the three parameter model. Stability of item parameter estimates was influenced more by differences in group ability than by the length of time between administrations. The item parameter estimates obtained for aptitude test data (SAT-V and SAT-M) had a higher degree of stability than those estimated for achievement tests. Items should be re-calibrated periodically to ascertain if parameter estimates have remained valid for a particular application and examinee population. (DWH)