Medication nonadherence in severe mental disorders is an important clinical issue, but estimates vary between studies. There is a need for valid self-reports for both research and clinical practice. This study examined the level of adherence to prescribed medication in outpatients with severe mental disorders and evaluated the validity of a simple self-report rating of adherence. From an ongoing study of severe mental disorders, 280 patients with schizophrenia and bipolar disorder who were prescribed psychopharmacological agents were included. We assessed adherence with serum concentration of medicines and tested the sensitivity and specificity of a simple self-report questionnaire for patients and compared with a report from health personnel. Adherence rate defined by serum concentrations within reference level was 61.6% in the total sample, 58.4% for schizophrenia and 66.3% for bipolar disorder. The patients' self-report scores overestimated adherence, but correlated significantly to health personnel scores (r = 0.50) and to serum concentration of medication (r = 0.52); the positive predictive value was 70%, and the negative predictive value was 91%. In this naturalistic sample, outpatients with severe mental disorders showed relatively good adherence to prescribed medication, and self-report questionnaires seem to be a valid method for measuring adherence.