A salivary mutans streptococci test and past caries experience were used as predictors for caries increment in a 3-yr study comprising 655 12-yr-old schoolchildren from two areas with contrasting levels of fluoride in the drinking water. The mean caries (DMFS) increment was similar in both groups during the study period, but a significantly (P < 0.05) higher incidence of approximal enamel lesions was registered in children from the high fluoride area. In both groups, a statistically significant (P < 0.05-0.001) positive relationship between salivary mutans streptococci score and/or past caries experience at baseline on one hand and caries increment during the study period on the other was established. The past caries experience was the most powerful predictor of caries risk in both the low fluoride and the high fluoride area. The sum of the sensitivity and specificity was somewhat higher in the low fluoride area (138%) compared to the high fluoride area (123%). The salivary bacterial enumeration used alone or in combination with past caries experience as well as past approximal caries experience were less useful as predictors in both groups. The present findings indicate that the natural fluoride exposure has a limited influence on caries risk assessment and the caries predictive ability of the salivary bacterial test and past caries in populations with a low level of disease.