The development of approximal carious lesions was monitored on annual bite-wing radiographs obtained from 315 children, participating in a controlled dentifrice trial; 157 of these children had used a dentifrice containing 250 ppm F and 158 a dentifrice containing 1,000 ppm F (both as NaF) during a 3-year period as the sole fluoride supplement. The total number of surfaces recorded was 6,099 at the baseline and 7,358 at the final examination. In the 250-ppm F group 393 approximal lesions (297 in the outer half, 96 in the inner half of the enamel) and in the 1,000-ppm F group 399 lesions (319 and 80, respectively) were diagnosed at the baseline examination. After 3 years the fraction of sound surfaces had decreased from 79% at baseline in both groups to 54% in the low-F group and to 58% in the high-F group. Concurrently, the relative number of enamel lesions had increased from 13% in both groups to 26% in the low-F group and 24% in the high-F group. 66% of the surfaces in the low-F group and 60% in the high-F group remained sound, and the relative proportion of static enamel lesions was 26 and 28%, respectively. The differences did not reach statistical significance. The results imply that higher fluoride levels in the dentifrice are more effective in retarding the initiation of caries, while they do not differ with respect to the progression of radiographically detectable approximal carious lesions.