• Bulimia nervosa;
  • outcome studies;
  • predictor of outcome;
  • research methods

The purpose of the study was to test if design and research methods of studies predict outcome results of bulimia nervosa. Fifty-seven articles on outcome of bulimia nervosa were found through MEDLINE and PSYCHINFO databases and 25 were included in the statistical analysis. Percentages of patients with a good outcome ranged from 24% to 74% with a mean of 51.1%. The variables best predicting outcome were time to follow-up and number of symptoms evaluated in definition of good outcome whereas dropout, design of studies, treatments, and sample characteristics did not predict significantly. The implications of these findings are discussed.