OBJECTIVES: Due to varying survey methods and samples, annual direct cost estimates for the care of major depression (MD) in Germany vary between 1264€ and 2577€ per patient in bottom-up cost analyses. To obtain more precise estimates for costs of MD, we drew a sample from claims data of a statutory health insurance covering approx. 11% of the German population. By accounting for levels of disease severity and comorbidities, we (a) provide a stratified cost analysis for MD in Germany and (b) explore potential bias caused by small samples or population selection criteria. METHODS: We selected a patient cohort aged 18 to 65 and not treated for depression in the previous 6 months with (a) at least two secured diagnoses of depression in 2010 and (b) the first depression-related service utilization between 01.01.2010 and 15.02.2010. Depression-related costs for outpatient-care, inpatient-care, and antidepressants were analyzed for the remaining year 2010 in total and by provider. Patients were stratified into three disease severity (ds) levels: minor/unspecified/atypical (ds1), moderate (ds2), and severe (ds3) MD. Furthermore 128 (co)morbidity groups were analyzed. RESULTS: The sample comprised 18,139 patients with a mean of 887€ in annual direct costs for care of MD. Mean costs by disease severity and mental comorbidity (mc) were for (ds1): n = 11,978 (66%), 623€ (without/with mc 509€/737€); for (ds2): n = 4,420 (24%), 1231€ (without/with mc 1056€/1601€); for (ds3): n = 1741 (10%), 1838€ (without/with mc 1389€/2483€). While mental comorbidities significantly explained higher direct costs for depression for every provider, somatic comorbidities did not. CONCLUSIONS: Our sample revealed (a) a large proportion of patients with low disease severity (ds1) and (b) large cost differences explained by disease severity and mental comorbidities. Comparable studies reporting higher mean costs for depression may possibly underrepresent patients with low disease severity in their sample.