A study of mortality for all patients with a first hospital diagnosis of schizophrenia in Stockholm County, Sweden, during 1973 to 1995 was performed, by linking the in-patient register with the national cause-of-death register. Overall and cause-specific standardized mortality ratios (SMR) were calculated by 5-year age classes and 5-year calendar time periods. The number of excess deaths was calculated by reducing the observed number of deaths by those expected. Our results confirmed a marked increase in mortality in schizophrenia both in males and females. Natural (somatic) causes of death was the main cause of excess deaths, with more than half of the excess deaths in females, and almost half of the excess deaths in males. Suicide was the specific cause of the largest number of excess deaths in males, while in females it was cardiovascular disease. SMRs were increased in both natural and unnatural causes of death, with 2.8 for males and 2.4 for females for all deaths, but were highest in suicide with 15.7 for males and 19.7 for females, and in unspecified violence with 11.7 for males and 9.9 for females. SMRs in suicide were especially high in young patients in the first year after the first diagnosis.