Purpose – The purpose of this paper is to examine household characteristics and schooling decisions in terms of enrollment and type of school in an urban setting in Nairobi.

Design/methodology/approach – The paper uses a cross-sectional data set collected in 2005. The sample comprises 7,475 primary school-aged children. A probit model was estimated to show what influences decisions at household level.

Findings – Analysis shows that different household and individual attributes motivate different decisions. A considerable proportion (40 per cent) of children from the poorest quintile attends non-public schools compared to 34 per cent from the richest quintile. The findings reveal that better-off households are more represented in the free primary education (FPE) programme. The predicted probability of a decision to attend a public school for a primary school-age child increases as the household wealth increases.

Practical implications – The paper concludes that poorer households are least attending and may be excluded from free public schools.

Originality/value – The paper demystifies the notion that introduction of FPE in developing countries is a pro-poor policy.