Despite the widespread influence of the alcohol dependence syndrome concept on the major nosological classification systems, little work has been done to test the validity of the alcohol dependence syndrome in community samples. In addition, numerous questions have been asked about the validity of current definitions of alcohol abuse. We examined the cross-sectional validity of DSM-IV alcohol dependence and abuse in 936 household residents randomly selected and screened for elevated drinking. We investigated validity by testing the association of a set of seven "criterion" variables, external to the alcohol diagnostic criteria, with dependence and abuse diagnoses. Results indicated that dependence diagnoses were significantly associated with all criterion variables when compared to those with no diagnosis, even though all subjects had elevated drinking and the cases of alcohol dependence were mild. In contrast, abuse diagnoses did not show a pattern of association with the criterion variables when compared to no diagnosis. When associations were tested comparing dependence cases to those with abuse only, results were mixed. This study is one in a series of investigations in this sample of household residents screened for elevated drinking levels.