The results presented here are from the follow-up of the cohort of workers ever employed at the Chapelcross site of British Nuclear Fuels plc (BNFL) between 1955 and 1995. The study cohort consists of 2628 workers, 2249 of whom were male, who were first employed at the plant before 1 January 1996, and who have 63967 person-years of follow-up. The mean follow-up period is 24.3 years. The 2209 members of the cohort (84%) classified as radiation workers accumulated 185.1 person-sieverts of external radiation; their median cumulative dose was 39.1 mSv, and 95% of their cumulative doses were less than 339.3 mSv. The Chapelcross workers show the usual 'healthy worker' effect. To the end of 1995, there were 528 deaths among the total cohort (20%), including 449 (20%) amongst the radiation workers. When the dose was unlagged, a statistically significant association was noted between cancer registrations of the buccal cavity and pharynx and dose, based on five cases. When the dose was lagged by 10 years, a statistically significant excess relative risk was noted between all cancer morbidity and dose, 1.80 Sv(-1) (0.03 to 4.45), based on 162 cases. This result is driven by the non-significant, but high excess relative risk estimates from the 12 prostatic cancer registrations. A statistically significant association is noted between the eight deaths amongst radiation workers who had prostatic cancer as the underlying cause of mortality and cumulative external radiation dose when the dose was lagged by 0, 2 and 10 years. The association is unlikely to be causal. The finding has little biological plausibility as the strength of the association weakened as the dose lagging increased; it was strongest when the dose was unlagged and disappeared when the dose was lagged by 20 years. None of the workers who was registered for or died from prostatic cancer had ever been monitored for exposure to tritium or to 51Cr, 59Fe, 60Co or 65Zn. There is no evidence to date amongst the Chapelcross cohort of increased risk for cancers considered to be radiogenic based on studies of populations exposed to high levels of radiation.