The current evidence on the role of the social, behavioral, and community determinants of dental caries is based on cross-sectional analyses. The available evidence has not been based on analysis of multiple determinants within the same population. This longitudinal study addresses both of these limitations. The study included data from 788 dyads of children and their caregivers (77% follow-up), who were examined in 2002-03 and 2004-05. The families were assessed by calibrated dentists for severity of caries at both time periods. Additionally, the caregivers answered questionnaires administered by trained interviewers. The caries increment was adjusted for reversals. Significant predictors of higher caries increment were higher consumption of soda drinks, older age of child, greater weight-for-age, fewer dental treatment visits, higher baseline caries levels of children and their caregivers, dental fatalism, and neighborhood disadvantage status. The selected risk factors in the final model explained about 20% of the variation in the increment.