BACKGROUND: Existing diverse bottom-up estimations of direct costs associated with depression in Germany motivated a detailed patient-level analysis of depression-related treatment (DRT), -costs (DRC) and Comorbidity. METHODS: A large sickness fund's claims data was used to retrospectively identify patients aged 18-65 years with new-onset depression treatment between January 1st and February 15th 2010, and follow them until December 31st 2010, describe DRT, estimate associated DRC, and predict DRC with a generalised linear model. RESULTS: A total of 18,139 patients were analysed. Mean direct DRC were €783. Predictors of DRC regarding psychiatric comorbidities were: "Delusion, psychotic disorders and personality disorders" (DRC-ratio 1.72), "Alcohol/drug addiction" (1.82), "abuse of alcohol/drugs" (1.57). Predictors of DRC regarding medical comorbidities were: "Rheumatoid arthritis" (0.77), "atherosclerosis" (0.65), "pregnancy" (0.66), and "Osteoarthritis" (1.87). Of all patients, 60.8% received their most intense/specialised DRT from a general practitioner, a medical specialist (23.7%), a psychotherapist (8.0%), a medical specialist and psychotherapist (2.9%), or in hospital (4.6%). Serious psychiatric comorbidity nearly tripled depression-related hospitalisation rates. LIMITATIONS: Seasonal affective disorder and missing psychiatric outpatient clinic data must be considered. CONCLUSIONS: Estimated DRC are significantly below the assessment of the German national guideline. Differing definitions of observation period and cost attribution might explain differing German DRC results. Signs of hospital psychiatric comorbidity bias indicate overestimation of hospital DRC. Identified associations of DRC with certain medical diseases in older adults warrant further research. Up to one quarter of patients with severe depression diagnosis might lack specialist treatment.