Background: To examine the benefits of physical activity (PA) on diseases with a long developmental period, it is important to determine reliability of long-term PA recall. Methods. We investigated 15-year reproducibility of PA recall. Participants were 3605 White and African-American adults in the Coronary Artery Risk Development in Young Adults study, aged 33-45 at the time of recall assessment. Categorical questions assessed PA before and during high school (HS) and overall PA level at Baseline, with the same timeframes recalled 15years later. Moderate- and vigorous-intensity scores were calculated from reported months of participation in specific activities. Results: HS PA recall had higher reproducibility than overall PA recall (weighted kappa=0.43 vs. 0.21). Correlations between 15-year recall and Baseline reports of PA were r=0.29 for moderate-intensity scores, and r=0.50 for vigorous-intensity. Recall of vigorous activities had higher reproducibility than moderate-intensity activities. Regardless of number of months originally reported for specific activities, most participants recalled either no activity or activity during all 12months. Conclusion: PA recall from the distant past is moderately reproducible, but poor at the individual level, among young and middle aged adults.