Heightened scholarly interest in stability and change in criminal behavior has increased the demand for longitudinal data. One method that may enhance the quality of retrospective self-reported data, especially reports of timing, is the life event calendar (LEC). Using a sample of incarcerated women, we assess the validity of LEC measures of self-reported prevalence, frequency and timing of arrests over a three-year period. We also examine the validity of self--reported frequency and timing of arrest data by respondent and arrest characteristics. Results suggest that the LEC elicits valid data on prevalence and frequency of arrests, while self-reported timing of arrests is recalled with less accuracy. Saliency appears more relevant for the accuracy of self-reported frequency, as compared to timing, and substance use has no effect on validity. We discuss future research using the LEC, especially with regard to improving the recall of the timing and sequencing of criminal events. (Contains 5 tables, 1 figure, and 22 notes.)