Public preschool boosts academic skills in kindergarten, but little is known about whether that boost lasts to third grade because many studies stop directly assessing children after kindergarten. The current study tests for sustained associations between preschool attendance and an array of repeatedly measured, directly assessed language and math skills; we do this separately for public pre-K and Head Start, the two major publicly funded preschool programs. We draw on a large, racially diverse sample of children from families with low incomes in Tulsa, OK (N = 689, M[subscript age at 3rd] = 8.5 years). Using propensity score weighting, we compare children who attended school-based pre-K or Head Start to those who did not attend preschool. Both school-based pre-K and Head Start attenders outperformed preschool nonattenders on numeracy in third grade. There was weaker evidence of a sustained preschool advantage on language and literacy skills, and no evidence that associations differed by preschool program.