Background: The recent estimate of around 2 million HIV-infected people in Ethiopia derives from data that are sparse, especially in the rural areas where the majority (. 85%) of the population lives. We assessed HIV prevalence in almost 72 000 army recruits who resided in urban and rural areas prior to recruitment. Methods: Rapid HIV tests, HIVSPOT and Determine, were conducted on blood samples drawn at enrolment from almost 10 000 urban recruits, in 1999, and 62 000 rural recruits, in 2000. Socio-demographic data from recruits were available. Results: In urban recruits, overall HIV prevalence was 7.2%, ranging from 4.3 to 10.5% depending on region. In rural recruits, overall HIV prevalence was 3.8%, but the majority were farmers (57%) and students (18%) with an HIV prevalence of 2.7% and 2.6%, respectively. (Higher) level of education in rural recruits was associated with HIV infection. Rural recruits of the Muslim religion were less likely to be HIV infected than recruits of the Orthodox Christian religion (odds ratio: 0.7; 95% confidence interval, 0.65–0.84). Urban and rural residents of Amhara region were at higher risk of HIV infection. Conclusion: Prevalence in both rural and urban army recruits is below previous estimates. Geographic distribution of HIV is uneven. The impact of religion, education, and region on HIV prevalence suggests avenues for targeting HIV prevention efforts in Ethiopia.