Although considerable progress has been made over the past 2 decades in the field of eating-disorder epidemiology, there is room for improvement. Eating disorder diagnoses for children, need assessment, prospective longitudinal studies, studies of natural cause, treatment dose-effect studies, and bias are all areas that need careful consideration. Clinicians should consider establishing a Cochrane database for eating disorders. The mortality studies reviewed in this article are dominated by small, sparse, unbalanced data sets, and publication bias and bias caused by loss to follow-up limits the validity of the reported findings. Some methodologic advice is given in the hope that some of the more obvious pitfalls can be avoided in future research.