Two thousand two hundred seventy-one deaths were recorded among 17,800 asbestos insulation workers observed from January 1, 1967-December 31, 1976. There was little increase in cancer deaths or of asbestosis in less than 15-19 years from onset of employment. In general, though, the period of clinical latency was 2-4 decades or more and there were important differences among the several asbestos-associated diseases. Lung cancer peaked at about 30-35 years from onset and asbestosis at 40-45 years. Each tended to decline in incidence afterwards. Pleural and peritoneal mesothelioma reached their highest incidence later than lung cancer, but the incidence did not decline. In studies of effects of asbestos exposure, it appears advantageous to observe for at least 35-40 years or more from onset of exposure and to analyze the experience in duration-from-onset categories. If this is not possible, only the very limited early effects will be identified and the full import of the exposure may not be appreciated.