This study aimed to make a longitudinal analysis of interexaminer calibration reproducibility in diagnosing dental caries in posterior teeth, by examiners without previous experience in epidemiological studies. A group of 11 inexperienced examiners underwent theoretical-practical training and calibration assessments, assisted by a standard examiner. An examiner who did not participate directly in the research selected 5-year-old children with and without caries. The D3 diagnostic threshold was used to evaluate dental caries, based on the World Health Organization (WHO) criteria. The initial calibration (baseline) was performed after the theoretical-practical training session, and consisted of examining 20 children; the second calibration occurred three months later, and involved evaluating another 18 children. The interexaminer agreement was obtained by kappa statistics, and by overall percentage agreement. The paired t-test was applied to compare the values for kappa means and overall percentage agreement between the time points studied. At baseline, the values for kappa (> 0.81) and overall percentage agreement (> 95.63%) were considered high. At the 3-month calibration assessment, all the examiners showed some decrease in both kappa (p < 0.0001) and overall percentage agreement (p = 0.0102). The calibration process currently proposed by the WHO is effective. However, reproducibility was not maintained over time for inexperienced examiners evaluating the posterior teeth of 5-year-old children, under epidemiological conditions.