With the introduction of free primary education (FPE) in Kenya in 2003, it was expected that the burden on poor households in financing primary education would be reduced substantially. This in turn would increase enrolment in public schools and lead to universal primary education. However, studies have shown that a considerable proportion of households in urban slums continue to enrol their children in pro-poor fee charging informal schools. The reasons presented in the available literature to explain this phenomenon of why poor slum-residing households bypass free public education are varied and some are simply speculative. In this paper, we hypothesise that poverty dynamics can partly explain households' decisions on the type of school in which to enrol their children. The analysis is based on longitudinal data collected by the African Population and Health Research Centre (APHRC) using urban demographic surveillance in two slums of Nairobi, Kenya. The data covers the period between 2005 and 2009 with a sample of 6965 pupils spread across 3763 households. Logistic regression methods are applied. The findings reveal that moving in and out of poverty can affect the type of schooling decision a household makes, with one quarter of those moving out of poverty shifting schools. The findings demonstrate that there is both willingness and ability to pay by the slum residents that is driving the utilisation of the private schools. The decisions are not random occurrences, but seem systematic and rational--parents want quality and affordability and a good number of those whose economic situation improved do not seem to believe the public schools, even under FPE, offers quality. The policy implication to be drawn from these findings is that the private schools for the poor should not simply be dismissed as "informal schools" because it seems they have some features which are attracting parents to choose them and leave the state system. Free primary education policy is being "rejected" by a good number of parents in the slums and this needs further investigation, as "excess" demand as suggested in some research papers does not seem to be the only explanation. (Contains 5 tables and 1 figure.)