SETTING: Laos, where the implementation of a national tuberculosis program started in 1995. OBJECTIVE: To estimate the prevalence (P) and annual risk of tuberculosis infection (ARTI) and relate this to case rates in the tuberculosis program. METHODS: Tuberculin survey in schools in Vientiane municipality and three provinces. Sampling was not representative, but included schools in accessible districts in provinces with different characteristics. A group of 150 tuberculosis patients were tested for comparison. RESULTS: An intra-dermal tuberculin test (PPD RT23/Tween80) was applied to 11237 children; 9968 reactions were read, 4035 among first and second grade children with no BCG scar: 818 in Vientiane (mean age 8.4), 803 in Savannakhet (8.9), 1563 in Xiengkhuang (8.4), and 851 in Luangprabang (7.9). If reactions of greater than or equal to 10 mm are counted, ARTI is estimated as 1.1% for the areas included in the survey with a strong variation by province (Vientiane 1.3%; Savannakhet 1.4%; Luangprabang 0.6%; Xiengkhuang 0.1%). The difference in ARTI is in accordance with differences in reported case rates in the provinces. The ARTI varied from 0.5% to 1.2%, depending on the method used for calculating P (Vientiane 0.6%-1.4%; Savannakhet 0.5-1.5%; Xiengkhuang 0.1%-0.2%; Luangprabang 0.4%-0.90%). CONCLUSIONS: The ARTI is unequally distributed geographically and/or by ethnic group. It is generally higher in south and lower in central and north Laos. Improved methods are needed for estimating the prevalence of infection.