To estimate the odds and prevalence of clinically relevant depression in adults with type 1 or type 2 diabetes. Depression is associated with hyperglycemia and an increased risk for diabetic complications; relief of depression is associated with improved glycemic control. A more accurate estimate of depression prevalence than what is currently available is needed to gauge the potential impact of depression management in diabetes. MEDLINE and PsycINFO databases and published references were used to identify studies that reported the prevalence of depression in diabetes. Prevalence was calculated as an aggregate mean weighted by the combined number of subjects in the included studies. We used chi(2) statistics and odds ratios (ORs) to assess the rate and likelihood of depression as a function of type of diabetes, sex, subject source, depression assessment method, and study design. A total of 42 eligible studies were identified; 20 (48%) included a nondiabetic comparison group. In the controlled studies, the odds of depression in the diabetic group were twice that of the nondiabetic comparison group (OR = 2.0, 95% CI 1.8-2.2) and did not differ by sex, type of diabetes, subject source, or assessment method. The prevalence of comorbid depression was significantly higher in diabetic women (28%) than in diabetic men (18%), in uncontrolled (30%) than in controlled studies (21%), in clinical (32%) than in community (20%) samples, and when assessed by self-report questionnaires (31%) than by standardized diagnostic interviews (11%). The presence of diabetes doubles the odds of comorbid depression. Prevalence estimates are affected by several clinical and methodological variables that do not affect the stability of the ORs.