The relation between total serum cholesterol and cancer incidence and mortality was studied in a cohort based on a probability sample of the United States population. 5125 men (yielding 459 incident cancers and 258 cancer deaths) and 7363 women (398 cases, 186 deaths) were initially examined in 1971-75 and followed up for a median of 10 years. Men in the lowest cholesterol quintile had nearly double the risk of those in the highest quintile for both incidence and mortality. Among women a similar relation was seen for cancer mortality, but cancer incidence in the lowest quintile was only 1.2 times that of women in the highest quintile. The inverse cholesterol-cancer relation in men was present for cholesterol determinations made 6 or more years before diagnosis of cancer. It may be premature to dismiss the inverse relation between serum cholesterol and cancer simply as a preclinical marker of disease.