Background: Caries risk assessment in preschool children is poorly validated in prospective studies. Aim: To validate the Bangkok checklist (BCL) in predicting caries development in a cohort of preschool children from low and moderate socioeconomic areas and compare it with two established risk assessment tools. Design: We followed 146 preschool children, aged 2-5 years for 2 years. At baseline, the caries risk category (low, moderate, high) was determined with three checklists: (a) BCL, (b) American Academy of Pediatric Dentistry form (CRAF), and (c) Caries Management By Risk Assessment (CAMBRA). Data were collected from questionnaires and clinical examinations. Caries increment was recorded by counting the number of surfaces that changed from sound to decayed. We used Kendall's tau, Poisson regression models and ROC analysis to assess the predictive ability of the different checklists. Results: Over 50% of the children developed new caries on a yearly basis. The BCL assigned the majority of the children (87%) into the high-risk category. Sensitivity was 88% but specificity was low (16%). Accuracy of BCL was similar to CRAF but inferior to CAMBRA. Conclusions: Bangkok checklist had a very limited performance in accurately predicting future early childhood caries in this population.