Using two different methods-bar and community household sampling-159 Latino gay men were recruited in the city of Tucson, Arizona. In addition to demographics, the study questionnaire assessed participants' sexual activity during the last 30 days with primary and nonprimary partners, condom use in the last year, and eight psychosocial constructs that have been predictive of HIV risk in different studies of (mostly white) gay/bisexual men. Questionnaires were available only in English; this Latino sample is thus likely to overrepresent highly acculturated, English-speaking men. Results show that 22% of the sample engaged in unprotected anal intercourse with nonmonogamous partners during the last 30 days; 51% of the sample reported at least one instance of unprotected anal intercourse during the last year. Of those men who practiced any anal intercourse during the last 30 days, 67% practiced unprotected anal intercourse with primary partners and 44% practiced unprotected intercourse with casual partners. Thus, the majority of Latino gay men who practiced anal intercourse in the month prior to the interview were not using condoms. Men who practiced unprotected intercourse with nonmonogamous partners reported lower annual incomes and were less educated. Two cognitive variables (behavioral intentions and perceptions of self-efficacy and self-control) and two behavioral variables (sex under the influence of alcohol and/or drugs and sex in public environments) emerged as the most important correlates of HIV risk.