The aim of this study was to assess the reliability of the Nyvad visual-tactile caries-diagnostic criteria when used among children who have been lifelong residents in areas with ‘optimal’ or low concentrations of fluoride in the drinking water. In each of two areas with drinking water fluoride concentrations of 0.3 and 1.1 ppm (0.3 and 1.1 mg/l) fluoride, respectively, 150 children were clinically examined twice, 2 wk apart, for dental fluorosis, using the Thylstrup-Fejerskov index (TF index), and for dental caries using the Nyvad visual-tactile caries criteria. The prevalence of dental fluorosis was 45% in the 1.1 ppm fluoride area and 21% in the 0.3 ppm fluoride area. When the results of the duplicate caries recordings were compared at the surface level, only minute differences were observed in the percentage agreement (91.7 and 90.7%, respectively) and in the kappa values (0.73 and 0.72, respectively). When individual DFS counts were compared across examinations using Bland-Altman plots and estimation of prediction intervals for the differences, we observed a greater variability of the differences between recordings among children from the low-fluoride area. Contrary to our expectations, a pronounced dental fluorosis background did not reduce the reliability of the caries recordings, which appeared to be slightly less reliable at very low levels of dental fluorosis.