The aim of the study was to evaluate the capability of caries-related salivary tests and a test based on past caries experience (baseline DFS) to select persons at high risk for caries. The subjects (n = 122) were 12-17 years old at the beginning of the study. Caries was registered and salivary samples were taken annually during the 3-study period. The 3-year caries increment was positively correlated to the baseline DFS (r = 0.46, p less than 0.001), salivary level of mutans streptococci (r = 0.30, p less than 0.001) and lactobacilli (r = 0.30, p less than 0.001), and combined level of mutans streptococci and lactobacilli (r = 0.39, p less than 0.001) and negatively correlated to the buffering capacity of saliva (r = -0.22, p less than 0.05). Tests based on either past caries experience or mutans streptococci or lactobacilli levels alone were not efficient in selecting persons at high risk for caries. Among the tests, DFS was the most sensitive and specific. A combination of either microbial test and DFS was more efficient to select persons at risk than various alternatives alone. The sensitivity was 84% and the specificity 62% for the combination of lactobacilli test and DFS and 71% and 79% respectively for the combination of mutans streptococci and DFS. In the former combination the positive prediction value was 43% and in the latter 56%.