The 1916 painters and the 1948 electricians who resided in the Canton of Geneva at the time of the 1970 census were identified and followed up to 1984. During the study period 121 disability pensions were awarded to painters and 59 to electricians. Age standardised incidence of disability per 1000 man-years at risk was higher among painters than among electricians for all neuropsychiatric causes (1.23/1000 and 0.68/1000, respectively) and for all other causes (5.50/1000 and 3.41/1000, respectively). No case of presenile dementia was diagnosed among painters. There was inadequate evidence to indicate that the higher risk of neuropsychiatric disability for painters might have been due to their occupational exposure to organic solvents. A possible toxic effect of these substances on the central nervous system was confounded with alcoholism which was associated with disability from neuropsychiatric disease in 12 of 20 painters and in only one of 10 electricians. Mortality and incidence of cancer were assessed among both cohorts and compared with the expected figures calculated from Geneva rates. Among painters there was a significant increase in overall mortality (O = 254, E = 218.5), in mortality from all cancers (O = 96, E = 75.4), and in incidence from all cancers (O = 159, E = 132.0). For the specific cancer sites, there was a significant excess risk for lung cancer (mortality: O = 40, E = 23.0), which was possibly related to occupational exposure to asbestos and to zinc chromate, although cigarette smoking was not controlled. The significant excesses of biliary tract cancer and of bladder cancer were in accordance with previous observations among painters from other countries. There was also a significant increase in incidence from testicular cancer (O=5, E=1.6), which has not been reported before. For causes of death other than cancer the excesses for alcoholism (O=5, E=0.8). for liver cirrhosis (O=14, E=8.8), for motor vehicle accidents (O=12, E=5.9), and for cerebrovascular disease when allowing for ten years of latency (O=8, E=4.0), were consistent with a probable increased risk of alcohol abuse. Among electricians overall mortality was similar to that expected (O=137, E=139.0). No significant excess risk was found for all cancers or for any specific cancer site. Because of the small number of expected deaths the statistical power was low for the assessment of a possible risk for leukaemia or for brain tumour.