BACKGROUND: Comorbid depression in the medically ill is clinically important. Admission to a general hospital offers an opportunity to identify and initiate treatment for depression. However, we first need to know how common depression is in general hospital inpatients. We aimed to address this question by systematically reviewing the relevant literature. METHODS: We reviewed published prevalence studies in any language which had used diagnostic interviews of general hospital inpatients and met basic methodological quality criteria. We focussed on interview-based studies in order to estimate the proportion of patients with a diagnosis of depressive illness. RESULTS: Of 158 relevant articles, 65 (41%) describing 60 separate studies met our inclusion criteria. The 31 studies that focussed on general medical and surgical inpatients reported prevalence estimates ranging from 5% to 34%. There was substantial, highly statistically significant, heterogeneity between studies which was not materially explained by the covariates we were able to consider. The average of the reported prevalences was 12% (95% CI 10-15), with a 95% prediction interval of 4-32%. The remaining 29 studies, of a variety of specific clinical populations, are described. CONCLUSIONS: The available evidence suggests a likely prevalence high enough to make it worthwhile screening hospital inpatients for depression and initiating treatment where appropriate. Further, higher quality, research is needed to clarify the prevalence of depression in specific settings and to further explore the reasons for the observed heterogeneity in estimates.