This paper describes DSM-IV and ICD-10 alcohol dependence prevalence rates and sociodemographic and drinking correlates. The sample under analysis (n = 2058) constitutes a multicluster probability sample of the US adult household population. The study response rate is 71%. The prevalence rate for current (past 12 month) DSM-IV alcohol dependence is 3.9%, and for current ICD-10 it is 5.5%. Agreement between DSM-IV and ICD-10 on whether respondents are dependent or not is less than optimal (Kappa = 0.67). The predictors of ICD-10 alcohol dependence are the frequency of drinking five or more drinks on occasion and age (inverse relationship). For DSM-IV alcohol dependence the correlates are drinking five or more drinks on occasion, being unemployed and age (also an inverse relationship). Differences in results underline the importance of understanding the variations among DSM-IV and ICD-10 criteria for alcohol dependence and the implications of these differences for epidemiological research. The high prevalence of dependence among young men may be the result of recognizing consequences of episodic heavy drinking as signs of alcohol dependence.