Occurrence and characteristics of depression and illness behavior were studied in 325 inpatients of a general hospital in the northern part of Italy. Patients were surveyed in six separate wards (medicine, surgery, dermatology, obstetrics and gynecology, orthopedics and ophthalmology) and represented about 90% of their actual population during a 1-week period. Two self-report scales were used for screening: the CES-D for measuring depression developed by the NIMH and the Illness Behavior Questionnaire (IBQ) developed by Pilowsky and Spence. Both scales were administered in their validated Italian translations. Applying the generally used cut-off point of 16 in the CES-D score, about 58% of the patients were classified as depressed. This cut-off was derived from community studies in the US. Applying a rather more conservative cut-off point of 23 (derived from the scores of psychiatrically depressed patients in Italy), still 33.5% of the patients emerged as depressed. The IBQ scores of the depressed patients showed significantly (p less than 0.001) higher levels of general hypochondriasis, disease conviction, dysphoria and irritability than the nondepressed patients. There were no relevant differences between wards in the amount of depression and IBQ scores even when differences were adjusted for age, sex, marital status and social class. Implications for psychosomatic research (sociodemographic characteristics of depression and illness behavior, bias in comparing hospital patients and controls in the general population, etc.) and treatment (consultation-liaison psychiatry) are discussed.