The relation between alcohol consumption and several causes of death, including breast cancer, was examined in a population of 581, 321 women enrolled in a prospective study in 1959 and followed for 12 years. Women who drank occasionally had about the same breast cancer mortality rate as nondrinkers; those who drank one to four drinks per day had SMRs 7-26% higher; five drinks per day, 1.89; and six or more drinks per day, 1.65. The two highest-consumption groups' risks were significantly higher than those of nondrinkers after multivariate adjustment for several breast cancer risk factors. Distinctive dose-response relationships were observed for two known alcohol-related conditions: cirrhosis of the liver and cancer of the aero-digestive tract, suggesting that results for other causes are not seriously biased by misclassification of drinking habits. Death rates from all causes combined were elevated for drinkers of three or more drinks per day. Whether or not the association of elevated breast cancer death rates ultimately turns out to be causal, there is ample reason to continue to warn the public against excessive drinking.