The aim of this study was to evaluate the influence of initial noncavitated caries lesions on cavitated caries increment in preschool children. A 2-year cohort study was designed to include the participants of a survey on dental caries performed in 2010. Preschool children (12-59 months old) were examined for dental caries and classified as children with no caries lesions, with only initial lesions, with at least 1 moderate caries lesion, and with extensive lesions. Socioeconomic data were also collected. After 2 years, 466 children were re-examined (follow-up rate of 72.9%) only for cavitated lesions. Association between caries incidence at 2 levels of severity and caries experience and other variables was evaluated using hierarchical Poisson regression analysis. The children with moderate and extensive caries lesions at baseline presented a higher risk of presenting both outcomes than the children with no caries lesions. Nevertheless, the children with only initial lesions had a higher risk of developing at least 1 new cavitated carious lesion, but not for a more severe increment in caries. Subgroup analysis stratified by the children's age showed that the influence of the presence of initial caries lesions on cavitated caries increment was only observed in children aged 12-35 months. In conclusion, although the presence of moderate and extensive lesions at baseline is a significant predictor for cavitated caries increment after 2 years in preschool children at all ages, the presence of only initial caries lesions is also associated, but with less severe caries incidence.