Incidence of cancer and mortality were studied among 674 men exposed to mercury vapour for more than one year at two chloralkali plants. Mercury excretion in urine had been monitored among the workers at the two plants since 1948 and 1949. An individual cumulative urinary mercury dose was calculated, based on about 20,000 urinary mercury measurements. The incidence of cancer and the mortality were followed up from 1953 to 1989 and 1953 to 1988 respectively. The general Norwegian male population served as a reference population. There was a lung cancer excess of borderline significance (standardised incidence ratio = 1.66, 95% confidence interval = 1.00-2.59). The introduction of a 10 year latent period before developing lung cancer did not increase the incidence ratio. The excess may be partly explained by the smoking habits in the cohort or possibly by exposure to asbestos. No excess of cancer was found in the target organs for mercury toxicity--namely, the kidney and the nervous system. No significant excess mortality was found for nephritis and nephrosis or nonmalignant diseases of the nervous system.