In a cohort of 4,563 nuclear workers followed retrospectively from 1950 to 1994, we found that age at exposure modified the effects of external radiation dose on cancer mortality. Analyses involved application of conditional logistic regression to risk sets of age- and calendar time-matched cancer deaths, with covariates treated as time dependent and with cumulative radiation doses divided according to the age intervals in which exposure occurred. After adjustment for confounding factors, we found that workers exposed to external radiation after the age of 50 years experienced exposure-related elevations in mortality from cancer at any site [rate ratio (RR) = 1.98; 95% confidence interval (CI) = 0.63-6.26], radiosensitive solid cancer (RR = 3.29; 95% CI = 1.10-9.89), and lung cancer (RR = 3.89; 95% CI = 1.23-12.3) substantially greater (1.6- to 3.5-fold greater) than were seen in coworkers exposed at all earlier ages. In contrast, all of the radiation doses contributing to mortality from cancers of the blood and lymph system were received before age 50 (for age <50, RR = 2.73 and 95% CI = 1.46-5.10; for age > or =50, RR = 0.24 and 95% CI = 0.00-687). Our results for cancer of any site are consistent with the results of previous studies examining the effects of exposure age in nuclear workers. Thus, effects of low-level radiation doses may depend on exposure age, and furthermore, patterns of effect modification by age may differ by type of cancer.