Abstract Nine teachers of clinical dentistry were presented with 12 pairs of duplicated bitewings on four different occasions. At the first two examinations they indicated the presence of caries in need of treatment and at the second two the degree of caries. The diagnostic sessions were repeated to measure reproducibility. A norm for the validity of degree of enamel caries and for the estimation of specificity and sensitivity of the diagnostic categories was applied. The teachers saw on average less caries and less deep caries than the norm. Variability by the teachers was especially large between examinations for enamel lesion categories. There was a mean of 69% agreement by the teachers with their own treatment indication criteria. Over recording of treatment need (mean 19%, SD 20%) and under recording (mean 36%, SD 17%) occurred at the same viewing periods. Using Cohen's kappa, it was found that despite their variation and inaccuracy this group of teachers should select caries in dentin as the teaching criterion for treatment need. It is suggested that the sensitivity and specificity with which dental practitioners can use a criterion should be taken into account firstly when selecting it, and secondly when choosing the frequency of radiographic screening.