This investigation explored the prevalence and predictive value of childhood obsessive-compulsive personality traits (OCPTs) in the development of eating disorders (EDs) using a novel retrospective questionnaire. To reduce bias associated with retrospective self-report data, an identical informant version of the questionnaire was also utilised. Substantial test-retest and inter-rater reliabilities were found for the questionnaire, as well as concordant validity with the semi-structured interview from which it was derived. Participants with an ED (n = 246) endorsed more childhood behaviours reflecting OCPTs than the control group (n = 89). This was mirrored in the informant report data (n = 93). The prevalence rate for each OCPT in childhood was significantly higher in the total ED sample compared to the control group. Both proband and informant reports of childhood traits predict the later development of an ED according to a strong dose-response relationship. The potential utility of this measure in future retrospective and prospective research studies is highlighted.