We previously reported a high degree of seasonality as measured by the Seasonal Pattern Assessment Questionnaire (SPAQ) in 41 patients with eating disorders (ED) compared with control subjects and patients with five other affective spectrum disorders. To clarify the relationship of the specific ED diagnosis and latitude to seasonal variation in a larger sample, we administered the SPAQ to 159 women with ED as defined by DSM-III-R. Subtype diagnoses were as follows: bulimia nervosa (BN), n = 109; anorexia nervosa (AN), n = 30; BN+AN, n = 20. Patients were studied at three locations: National Institute of Mental Health (NIMH), n = 46; Medical University of South Carolina (MUSC), n = 53; University of Michigan (UM), n = 60. The control group comprised 50 female normal volunteers studied at NIMH. There was a statistically significant difference in Global Seasonality Scale (GSS) scores among the four diagnostic groups, and all ED subtypes had significantly higher GSS scores than control subjects after post hoc Bonferroni t tests. Higher GSS scores were also found in patients with BN+AN compared with patients with BN or AN alone. The patterns of change were similar to those observed in winter seasonal affective disorder (SAD). Thirteen percent of the total sample of ED patients met SPAQ criteria for winter SAD, with 2.5% each for summer SAD and subsyndromal SAD. UM BN patients (latitude 42 degrees N) had higher GSS scores and a higher prevalence of winter SAD than MUSC BN patients (latitude 33 degrees N), but this difference was not statistically significant. These data support the hypothesis that ED and SAD may involve similar pathophysiological mechanisms, possibly related to serotonin dysregulation.