Evidence that chess expertise rests on practice alone mostly comes from studies using a correlational retrospective recall paradigm, which confounds amount of study with number of games played and possible innate talent. Researchers also often use latest performance rating and include participants who play and study little. Study 1 partially replicated such studies with improvements such as use of peak rating and a large, skilled sample. Number of internationally-rated games played was the strongest predictor of peak rating. Total study hours was a significant but weaker predictor. Study 2 controlled for sampling confounds by including only very well-practiced players who had played at least 350 internationally-rated games. Total study hours did not predict rating at 350 games. Study 3 found that the subjective phenomenon of reaching a performance ceiling and undertaking specific practice to get beyond it does occur but does not distinguish between stronger and weaker players. Study 4 found that many players play relatively few internationally-rated games mostly because of other commitments, such as work and education. Extensive study may go along with great interest in and persistence at chess but apparently lacks a major causative role in chess performance level.