Only a longitudinal study, in which retrospective reports can be verified against data collected earlier, can determine what topics tend to be reported accurately and whether certain types of individuals are more likely to be accurate reporters. A representative group of adults who became parents 50 years ago are now part of the oldest and longest continuing longitudinal study of adulthood and aging. Subjects were first interviewed at mean age 29 and again at 47 and 69. Factual questions about such topics as birthplace, education, and occupation, and attitudinal variables such as bonds and relationships with spouse, children, and own parents, were analyzed for consistency over time. Replies given by subjects at each follow-up interview were compared with previously reported information. No sex differences were found for any comparison. No differences were found in background variables; neither education nor occupation predicted consistency in reports. Intelligence did not distinguish persons judged to be high or low in accuracy. No relationship was found between personality measures and accuracy of retrospections. Although subjects showed considerable consistency in their retrospective reports over 40 years of adulthood, factors which could account for individual differences over time or topics were not identified. The findings indicate the need for further investigation using other analytical methods such as case study. (Author/NRB)