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 sample of women who became mothers 50 years ago are now part of the oldest and longest continuing longitudinal study of adulthood and aging. Subjects (N=44) were first interviewed in 1928-1929, again in 1945-1947, and again in 1968-1969. The average age at the last interview was 67.9 years. Factual questions about such topics as birthplace, education, and occupations, and attitudinal variables such as bonds and relationships with spouse, children, and own parents, were examined for consistency over time. Subject responses at each follow-up interview were compared with earlier information. Factual variables were recalled with greater accuracy than attitudinal variables, and accuracy did not diminish with age. No subjects could be described as consistently low or high in accuracy over time; no discernable pattern emerged among the group. Personality characteristics showed considerable consistency over time. The results suggest that accuracy of recall does not appear to be a generalizable trait characteristic. (Author/NRB)