OBJECTIVES: High-quality preschools are known to prepare children for success in primary school. Over half of Indonesia’s children now pass through preschools whose quality and effectiveness are unknown. Our goal was to evaluate two government preschool models, namely kindergarten (TK) and the less formal health-post (PAUD), with and without capacity-building efforts of a non-governmental organization (NGO-Plan), on children’s language and math skills. METHODS: Thirteen TK and 17 PAUD Plan-supported and the same number of government-supported preschools were randomly selected from East Nusa Tenggara, Indonesia. Five children from each (n = 292) and five who had graduated from each and were now in first grade (n = 241) were randomly selected and tested on language and math measures. The Plan-supported preschools were assessed for quality. Mothers reported on their family’s socio-demographic situation and their child’s preventive health practices, illnesses and diet over the previous two weeks. RESULTS: Analyses of covariance adjusting for clusters indicated that children attending Plan-supported preschools performed better overall, and especially those in TK preschools. Plan-supported TKs were observed to have higher quality than Plan-supported PAUDs. First graders who graduated from Plansupported preschools, both TK and PAUD, achieved higher scores on language and math tests than government-supported graduates. Preventive health practices were better in the Plan group, though diet and height-for-age were poor overall. CONCLUSIONS: Upgrades to the government preschool program are needed to raise its quality and effectiveness, specifically by introducing a mix of instructional and indoor free-choice play, resources and teacher training to support children’s learning.