OBJECTIVE: Racial differences in variables that predict return to the psychiatric emergency room were examined. METHODS: A random sample of 319 clients was obtained from the logs of a psychiatric emergency room of a state-operated, acute care psychiatric hospital. The dependent variable was a return visit to the psychiatric emergency room within 18 months of the index visit. Separate logistic regression equations were calculated for African Americans (N = 163) and Caucasians (N = 156) to estimate the moderating effects of race. RESULTS: Four variables predicted return to the emergency room for both African Americans and Caucasians: previous visits to the psychiatric emergency room, previous psychiatric hospitalizations, current receipt of outpatient treatment, and nonreceipt of aftercare following the index visit to the emergency room. Three unique predictors were found for African Americans: never having been married, not living in stable housing, and not being admitted at the index visit. CONCLUSIONS: Generally, repeat visitors from both racial groups tended to be chronic users of psychiatric services who may be using the psychiatric emergency room for routine psychiatric care. However, race was also an important moderator variable; several risk factors predicted a return visit only for African Americans.