In developing countries, the prevalence of dental caries in children remains high, which means that implementing a simple and convenient classification is critical. The classification needs to be evidence-based and needs to reflect tooth-level information. In this study, the prevalence of dental caries in the primary dentition of 352 Myanmar school children at the ages of 5, 6, and 7 was analyzed at the tooth level to clarify the underlying data structure of the patterns of dental caries in the population. Ninety-three percent of subjects had caries in primary dentition and the mean number of decayed teeth in primary dentition was 7.54 ± 4.82. Based on the item response theory analysis, mixed-effect modeling, and Bayesian network analysis, we proposed the following classification: Group 1: No dental caries; Group 2: Dental caries in molar teeth or dental caries in maxillary anterior teeth; Group 3: Dental caries in both molar and maxillary anterior teeth; Group 4: Dental carries in mandibular anterior teeth. Dental caries (dmft) in the groups was different between groups. The results of characteristics of tooth-level information and classification presented in this study may be a useful instrument for the analysis of the data of dental caries prevalence in primary dentition.