Objectives Patients' self-report of medicine taking is a feasible method of assessing their adherence to prescribed pharmacological treatment. Aim of this study was to assess whether the German version of the Medication Adherence Report Scale (MARS-D) is an appropriate instrument for measuring patient adherence. Methods After translation into German, the questionnaire was sent to 1488 patients with chronic diseases and patients with risk factors of cardiovascular disease. Reliability and validity of the MARS-D were assessed and compared with the psychometric properties of the original English version. The relationship between patients' characteristics and adherent behaviour was estimated using bivariate correlation and a linear regression model. Results The MARS-D was analysed if patients were taking medicines and the MARS-D was complete leaving 523 (35.1%) analysable questionnaires. Internal consistency of the MARS-D (Cronbach's alpha 0.60-0.69) was satisfactory and comparable to the English original (Cronbach's alpha 0.69-0.90). Test-retest reliability was satisfactory (Pearson's r 0.61-0.63), however, lower than in the English sample (r = 0.97). Convergent validity was low but showed statistical significance. Patient socio-demographic characteristics had weak influence on MARS-D score indicating high reported adherence for older patients (P < 0.05), patients with German mother tongue (P < 0.05) and high number of medicines (P < 0.01). Conclusions Preliminary psychometric evaluation of the MARS-D is encouraging. MARS-D is an appropriate measure to detect patients at risk of non-adherence. The MARS-D could be used in routine care to support communication about the medication taking behaviour, as self-report of non-adherent behaviour corresponds to the facts.