Data from a national representative sample of US adults were analyzed to determine the association between the relative frequency of heavy drinking (the proportion of drinking occasions on which 5+ drinks were consumed) and past-year alcohol dependence, adjusting for the influences of average ethanol intake and sociodemographic factors. Fifty-seven percent of current drinkers reported never drinking 5+ drinks, and 21% drank 5+ drinks at least once but on less than 10% of all drinking occasions. Nine percent reported drinking 5+ drinks on at least half of all drinking occasions. Average daily intake was positively correlated with the relative frequency of heavy drinking, and both consumption measures were positively associated with the risk of alcohol dependence. Increases in either relative frequency of heavy drinking or average ethanol intake reduced, but did not eliminate, the effect of the other on the risk of dependence. The excess risk of dependence associated with frequent heavy drinking varied among population subgroups and was increased by age, education, and female gender.