A case-control analysis among 53 Portsmouth, New Hampshire, Naval Shipyard (SIC-3731) workers who died from leukemia, and 212 matched controls selected from a cohort of 24,545 workers employed at the shipyard was conducted in an attempt to ascertain associations between leukemia and specific exposures and occupations. The shipyard is a large industrial complex involved in designing, manufacturing, overhauling, and refueling of conventional and nuclear naval vessels. Radiation dose histories and detailed work histories by job and shop were evaluated for each worker studied. No statistically significant associations with leukemia were found among workers exposed to ionizing radiation. An increase in the odds ratio for myeloid leukemia was observed among persons who had accumulated a lifetime radiation dose of at least 1.0 roentgen equivalent man (rem). For workers exposed to organic solvents, the odds ratios were elevated for all of the leukemia cell types, but none of the odds ratios reached statistical significance. Statistically significant excess risks of leukemia were noted for electricians and welders. The odds ratio was significantly elevated for all leukemias in electricians, particularly for lymphatic leukemia. The odds ratio was significantly elevated for myeloid leukemias in welders, but not for all leukemias. Results were observed by univariate analysis and conditional logistic regression analysis.