Expertise typically develops slowly over years, and controlled experiments to study its development may be impractical. Researchers often use a correlational, retrospective recall method in which participants recall career data, sometimes over many years before. However, recall accuracy is uncertain. The present study investigated the accuracy of recalled career data for up to 38 years, in over 600 international chess players. Participants' estimates of their entry year into international chess, total career games played, and number of games in a typical year were compared with the known true values. Entry year typically was recalled fairly accurately, and accuracy did not diminish systematically with time since list entry from 10 years earlier to 25 or more years earlier. On average, games-count estimates were reasonably accurate. However, some participants were very inaccurate, and some were more inaccurate in their total-games counts and entry-year estimates. The retrospective recall method yields usable data but may have some accuracy problems. Possible remedies are outlined.