The reported milk consumption habits of 3,334 cancer patients and 1,300 comparable control subjects seen at Roswell Park Memorial Institute were studied. The cancer patient group included persons diagnosed with cancers of the oral cavity, stomach, colon, rectum, lung, breast, uterus, cervix, prostate, or bladder. Data were obtained by means of a standardized admissions questionnaire. As a group, control patients were more likely to report never drinking whole milk; cancer patients more often reported frequent consumption of whole milk. Relative risks of different patterns of milk consumption for specific cancer diagnoses, adjusted for sex, age, smoking history, education, and county of residence, were calculated with multiple logistic regression analysis. Elevated risks for frequent consumption of whole milk relative to not drinking milk were observed for cancers of the oral cavity, stomach, colon, rectum, lung, bladder, breast, and cervix. Reduced risks for frequent consumption of 2% milk relative to not drinking milk were observed for cancers of the oral cavity, stomach, rectum, lung, and cervix. Preference for exclusive consumption of reduced-fat milk was linked to significant risk reduction for oral and cervical cancers, and drinking only whole milk was linked to significant risk increases for cancers of the oral cavity, stomach, rectum, lung, and breast. Some associations were observed for a computed index of milk fat intake, but the overall pattern of effects was not fully explained by variations in fat content. The effects observed for some sites may be confounded by other dietary or nondietary correlates of risk.