Core respondents were interviewed as part of the 1995 National Alcohol Survey. This study used a multistage, multicluster, probability sampling frame, with oversamples of Blacks and Hispanics. Only married or cohabiting couples were included for the final stage selection. The analysis was limited to couples in which both partners described themselves as white (n=555), Black (n=358), or Hispanic (n=527), yielding a final sample of 1,440 couples. Participants were asked about occurrences of 11 physically violent behaviors during the past year that they may have perpetrated against their partners or that their partners may have perpetrated against them. Measures of SES were income, employment status, and education. IPV was measured with the Conflict Tactics Scale, Form R. For each racial/ethnic group, a series of t tests was performed to determine whether mean household income and mean years of education differed significantly between couples that did and did not report partner violence. Deviance statistics, based on contrasting transformed likelihood measures obtained through multivariate logistic regression models, were computed to assess the relative influence of SES on the probability of IPV. Findings show that annual household income had the greatest relative influence on the probability of partner violence, regardless of racial/ethnic group. The relative influence of SES on the probability of IPV varied across racial/ethnic groups and by perpetrator gender. Despite the study limitations noted, these findings show the importance of measuring and including SES indicators within IPV research. 3 tables and 49 references