We compared HIV prevalence, risk behaviours, and social and sexual norms among nine ethnic minority (Hilltribe) groups in northern Thailand. Communities were selected on the basis of size, ethnicity, development level and geography. Subjects (15-45 years) were stratified by gender and selected by household using two-stage randomization. Forty volunteers were identified in each of 27 villages. Participation was voluntary and informed consent was obtained. HIV infection status was determined using ELISA and Western Blot on saliva samples. Risks for HIV were measured with structured interviews using local languages. Overall HIV prevalence was 23/1080 (2.13%) with an equal male-female ratio. HIV prevalence rates were: Shan 8.75%, Akha 5.0%, Yao 5.0%, Thin 1.25%, Hmong 0.63% Lahu 0.63%, Lisu 0.63%, Karen 0, and Pa-Long O. Sex worker use was an HIV risk for men (p = 0.0001), but injecting drug use was not; for women, having been a sex worker was a significant HIV risk (p < 0.0001). HIV rates, social norms and sexual behaviour varied considerably among ethnic groups, as did attitudes toward commercial sex work and use. HIV prevention needs to target the Shan, Akha, and Yao communities, and to focus on reduction of brothel work and use. In Thailand, public sector responses to acquired immunodeficiency syndrome (AIDS) have focused almost exclusively on the majority low-lander population. A comparative study of human immunodeficiency virus (HIV) prevalence, risk behaviors, and sexual norms among 9 ethnic minorities in the 6 northernmost Thai provinces revealed substantial variation in these hilltribe communities. 540 men and 540 women 15-45 years old from 27 villages were interviewed and tested for HIV. Each ethnic group included (Shan, Karen, Hmong, Akha, Lahu, Lisu, Pa-Long, and Thin) has its own language, culture, and history. Overall, there were 23 cases (2.13%) of HIV; HIV prevalence was 2.04% among men and 2.22% among women. This rate ranged from a high of 8.75% among the Shan to 0 among the Karen and Pa-Long. HIV-positive women were found in 7 of the 9 ethnic groups, while infected men were found in only 3 groups. In women, HIV prevalence was highest (3.9%) in the 20-24 year age group; in men, it was highest (4.7%) in the 25-29 year age group. 10 of 12 HIV-positive women, compared to only 3 of 11 HIV-positive men, were aware of their serostatus before the survey. Among men, ever having had sex with a commercial sex worker was the only behavioral factor associated with HIV infection; among women, being a commercial sex worker was a significant risk factor. Only 7.8% of women and 15.0% of men had ever used condoms with their spouse. The discordant rates of HIV among men and women in this sample suggest that infection is occurring outside the indigenous areas, primarily as a result of commercial sex encounters.