The aim of this study was to develop and evaluate an instrument to detect the various forms of clinical eating disorder that exist among young women in the community. A subsidiary aim was to test an alternative instrument based directly on the DSM-III-R criteria for eating disorders. Participants were 285 young adult women listed in the case registers of two general practices. Receiver operating characteristic analysis showed that the new (statistically derived) instrument performed well and was better at case identification than the only other instrument currently available. In contrast, the instrument based directly on the DSM-III-R criteria performed poorly. It was not possible to assess the ability of the new instrument to detect cases of anorexia nervosa since there were no cases in the sample.