The aim was to assess how accurately some commonly used risk factors/risk markers (predictors) for caries development could identify children with and without approximal caries as judged from bitewing radiography. Two hundred and sixtyseven consecutive 5-year-old children from two Swedish cities participated. Three experienced dentists examined the children. The predictors were the overall dmfs (de-cayed, missing and filled surfaces) value (canines and molars), the number of occlusal dmfs, the frequency of intake of between-meal sugary products, visible plaque on free smooth surfaces of second primary molars, toothbrushing habits and (before bitewing examination) an overall judgement by the examining dentist. The mean dmfs value without bitewing examination was 0.40 (SD = 1.22). Twelve percent of the children had at least one dentin lesion and 33% at least one enamel lesion that were detected from bitewing examination only. The gain from adding bitewing examination to clinical examination amounted to a mean of 1.2 approximal enamel and/or dentin lesions. The ability to correctly identify children with approximal caries from the predictors was limited; sensitivity ranged from 0.27 to 0.75 and specificity ranged from 0.41 to 0.93. The single best predictor was the dentist's overall judgement with an average precision of 73%; average sensitivity for the presence of enamel and dentin lesions was 0.48 and for the presence of dentin lesions 0.66. The rest of the predictors added little to the predictive power. It is concluded that 33% of the 5-year-olds, representing a low caries prevalence population, benefited from bitewing examination. The ability to identify these children from the predictors was, however, limited. Copyright (c) 2005 S. Karger AG, Basel.