A common assumption is that outreach-based HIV counseling and testing services reach a clientele with a higher HIV seroprevalence than clinic-based counseling and testing. To examine this assumption, we analyzed Wisconsin's anonymous counseling and testing client records for 62,299 contacts (testing episodes) from 1992 to 1995. Bivariate analysis of counseling and testing service setting (outreach-based or clinic-based) and HIV test results suggested that outreach contacts were 23% (odds ratio [OR], 1.23; 95% confidence interval [95% CI], 1.0-1.5) more likely to test HIV-seropositive than clinic-based contacts. Relations between HIV test outcome and variables for client age, race, gender, previous testing history, mode of risk exposure, and region, as well as service setting, were examined by logistic regression. An inverted relation between service setting and seropositivity (OR, 0.65; 95% CI, 0.5-0.8) indicated that, within some subpopulations, outreach contacts were significantly less likely to test HIV-positive than clinic-based contacts. Analysis of interactions among the covariates identified race as a critical codeterminant in the relation between service setting and test outcome. These results support retargeting outreach services to enhance their overall effectiveness. Specific recommendations include the need for aggressive strategies to better "market" HIV counseling and testing to nonwhite populations, and to focus resources more selectively on gay/bisexual men of all races.