To evaluate further the relation between alcohol consumption and total mortality, we have carried out new Cox proportional hazards model analyses of 21,535 deaths through 2002 in the Kaiser Permanente study. This follow-up includes 2,618,523 person-years of observation, with a mean follow-up of 20.6 years. We adjusted for age, sex, ethnicity, body mass index, marital status, education, and smoking. New methodology was used to stratify light-moderate drinkers into groups felt more or less likely to include under-reporters. The analysis reconfirms that the relation of alcohol drinking to total mortality is J-shaped, with reduced risk (mainly because of less cardiovascular disease) for lighter drinkers and increased risk for persons reporting more than 3 drinks per day. Infrequent (occasional) drinkers have risk similar to that of lifelong abstainers, while former drinkers are at increased risk, especially for noncardiac death. The general shape of the relation of alcohol to mortality is similar for men and women. Age differences are substantial, with the apparent benefit from light-moderate drinking not seen before middle life. Our data indicate further that the apparent magnitude of benefit of lighter drinking is probably reduced by systematic underreporting.