A previous report that the leukocyte count was related to mortality from cancer was evaluated in two large groups of multiphasic health examinees in the San Francisco Bay area, California--a cohort of about 25,000 persons followed up for mortality and a cohort of about 160,000 persons followed up for cancer incidence between 1964 and 1980. The leukocyte count was related to mortality from all cancers, smoking-related cancers, and cancers that were not smoking-related, but it was not related to cancer mortality in nonsmokers. The leukocyte count showed an association with the incidence of smoking-related cancers that was only partially removed by analytical control for smoking. It was only slightly, if at all, related to the incidence of not-smoking-related cancers and to the incidence of all cancers among nonsmokers. The relation of the leukocyte count to cancer mortality appears to be due to its close association with cigarette smoking, which raises the incidence of certain cancers and can hasten death attributed to cancer.