We have found shortened survival in 200 schizophrenics, 100 manics, and 225 depressives by comparing them with the Iowa general population. The numbers of years of shortened survival were estimated from the survival curves of each diagnostic group by sex. Variables affecting shortened survival were diagnostic group, sex, age at admission, and pay status at admission. Excess causes of death were suicides, accidents, and infective and circulatory system diseases. The absolute mortality for deaths due to neoplasm in our sample was not significantly different than that in the Iowa general population. However, the proportional mortality from the same data showed a deficiency in deaths due to neoplasm in schizophrenia and affective disorders. Well-designed epidemiologic studies are needed to see if there is a real deficiency in deaths due to cancer in the major psychoses.