To test the effects of an HIV/AIDS education program. A quasi-experimental, nested cross-sectional design including baseline and 6-month follow-up surveys. Schools, stratified according to location, were randomly assigned to intervention (n = 6) or comparison conditions (n = 12). Public primary schools in Arusha and Kilimanjaro regions, Tanzania. A total of 2026 sixth and seventh grade pupils (average age, 14.0 years) participated at baseline (85%) and 1785 at follow-up. The program was designed to reduce children's risk of HIV infection and to improve their tolerance of and care for people with AIDS. Local teachers and health workers attended a 1-week training workshop before implementing the program over a 2–3-month period (averaging 20 school hours per class). Self-reported exposure to AIDS information, communication regarding AIDS; AIDS knowledge, attitudes towards people with AIDS, attitudes towards having sexual intercourse, subjective norms regarding sexual intercourse, and intention to engage in sexual intercourse. Following this program, intervention pupils reported significantly higher scores for the following outcome measures than pupils attending the comparison schools: AIDS information (13.1 versus 10.5; P= 0.0001), AIDS communication (10.9 versus 7.8; P= 0.0001) AIDS knowledge (14.5 versus 11.5; P= 0.0001), attitudes towards people with AIDS (9.0 versus 6.7; P= 0.0008), subjective norms (45.5 versus 43.9; P= 0.011), and intention (1.3 versus 1.4; P= 0.020). No program effect was seen for attitudes towards sexual intercourse (47.0 versus 46.3, P=0.44). These results indicate that it is feasible and effective to provide AIDS education for Tanzanian primary school children.