PURPOSE: Cancer mortality data allow assessing, at the same time, the risk of developing the disease and the quality of care provided to patients after the oncologic diagnosis. This study explores the risk of death caused by a single tumor site in a psychiatric population treated in a community-based psychiatric service. METHODS: All patients with an ICD-10 psychiatric diagnosis, seeking care in 1982-2006 (25 years), were included. Data were drawn from the South Verona Psychiatric Case Register (PCR). Mortality and cause of death were ascertained using different procedures and sources. Standardized mortality ratios (SMRs) were used to compare the observed number of deaths with the expected number using as reference a population in the Veneto region. RESULTS: Having been admitted to the hospital (SMR = 1.32), having a short interval from registration (1.52), having a diagnosis of alcoholism (2.03), and being a middle-aged male (1.83) were factors showing an increased risk of death from cancer. Increased SMRs were found for cancer of the oral cavity (22.93), lymphoma, leukemias, Hodgkin's lymphoma (8.01), and central nervous system (CNS) and cranial nerve tumors (4.75). The SMR decreased for stomach tumors (0.49). Patients with alcoholism (5.90 for larynx), affective disorders (20.00 for lymphomas), and personality disorders (28.00 for SNC) were found to be exposed to a high risk of cancer death in specific sites. CONCLUSIONS: Psychiatric patients showed different patterns of site-specific cancer mortality when compared with the general population. The 20-fold higher risk of dying from hematological neoplasms needs further investigation. Chronic use of phenothiazines could be involved in the relative protection from stomach and prostate cancer found in psychiatric patients