Mortality and cancer incidence were studied in the National Registry for Radiation Workers in, relative to earlier analyses, an enlarged cohort of 174 541 persons, with longer follow-up (to 2001) and, for the first time, cancer registration data. SMRs for all causes and all malignant neoplasms were 81 and 84 respectively, demonstrating a 'healthy worker effect'. Within the cohort, mortality and incidence from both leukaemia excluding CLL and the grouping of all malignant neoplasms excluding leukaemia increased to a statistically significant extent with increasing radiation dose. Estimates of the trend in risk with dose were similar to those for the Japanese A-bomb survivors, with 90% confidence intervals that excluded both risks more than 2-3 times greater than the A-bomb values and no raised risk. Some evidence of an increasing trend with dose in mortality from all circulatory diseases may, at least partly, be due to confounding by smoking. This analysis provides the most precise estimates to date of mortality and cancer risks following occupational radiation exposure and strengthens the evidence for raised risks from these exposures. The cancer risk estimates are consistent with values used to set radiation protection standards.