Triazine herbicides, used extensively in the United States, have not been assessed adequately for carcinogenicity in humans. This study evaluated the mortality experience during the period 1960-1986 of 2,683 men with definite or probable manufacturing exposure to triazine herbicides and 2,234 men with possible exposure to triazines. Standardized mortality ratios (SMRs) were computed as the observed numbers of deaths among study subjects divided by the numbers expected on the basis of general population mortality rates. Subjects with definite or probable exposure to triazines had a favorable mortality experience, compared with U.S. men [all causes, SMR = 72, 95% confidence interval (CI) = 58-89; all cancer, SMR = 85, CI = 46-142]. This group had an increase in deaths from non-Hodgkin's lymphoma (NHL) (3 observed/0.78 expected, SMR = 385, CI = 79-1,124). Two of the decedents with NHL had worked for less than 1 year in triazine-related jobs. Among the 2,234 subjects with possible triazine-related work, the mortality rate for all causes combined was similar to the rates of U.S. men. There was only one confirmed death from NHL. On balance, the results were consistent with previous investigations, which have found no convincing evidence of a causal link between triazines and cancer. However, the exposed cohorts were relatively young and had, on average, only 18 years of follow-up. In particular, results were imprecise for subjects having both a long duration of exposure and long potential induction periods.