A total of 217 adult male subjects were evaluated utilizing event-related potentials (ERPs) elicited with two different auditory tasks (Counting and Choice Reaction). Ninety-eight alcoholics from high-density, multigenerational families were evaluated along with 39 first-degree nonalcoholic relatives from the same high-density families. Eighty controls, selected for low density of alcoholism in their extended families, were also studied. Using both conventional and topographic analyses, no significant differences in the amplitude of the P300 component could be found with either of the auditory tasks. No significant differences in amplitude of N250 were seen. The latency of N250 increased with increasing conditional probabilities (0.33, 0.67, and 1.00), a trend that was amplified in the Counting task as compared with the Choice Reaction task. This prolongation in a task not requiring a reaction response (button press) tended to increase the latency more for alcoholics than controls or high-risk nonalcoholic subjects. Age, lifetime, and recent drinking were treated as covariates in all analyses. The absence of P300 amplitude differences between adult high- and low-risk subjects is discussed in the context of the much more reliable differences seen between high- and low-risk children from the same high- and low-density families, when evaluated with the same auditory tasks.