In the Puerto Rico Heart Health Program, a prospective epidemiologic study, 9824 men living in defined urban and rural areas of Puerto Rico have been followed for approximately eight years. Periodic examinations have been used to assess potential cardiovascular risk factors and to monitor coronary heart disease events. Although the emphasis of the study has been oriented to cardiovascular disease, all causes of death have been closely monitored. During the observation period, 179 deaths from cancer have occurred in the age group 45--64 years. The distribution of cancer deaths varies from that of the mainland, with a greater percentage being oral, pharyngeal and stomach, and a lesser frequency being lung. Serum cholesterol measured at the first examination in 1965 was found to vary inversely with subsequent mortality from cancer. This trend was significant overall for the rural area and in the age group 45--54 years in the urban area. Multivariate analysis with other covariables did not eliminate the cholesterol relationship in the rural men, although relative weight, ventricular rate, hematocrit and cigarette smoking made independent contributions in certain age groups in both locations. The association between serum cholesterol and cancer mortality in the rural area and in the young urban area suggests that further investigation is needed to determine whether the relationship is causal, a secondary association to some other unidentified set of primary factors, or merely a physiological reaction to early and undiagnosed stages of the cancer process.