Using conventional benefit incidence analysis this paper analyses the distribution of public education expenditure in Bangladesh in 2005. The paper shows that public recurrent education expenditure is not pro-poor. Even at the primary level 47% of total public recurrent spending goes to the poor whereas 50% of all primary school aged children are poor. The paper also explores spending patterns of the primary and secondary education stipend programmes. At the primary level, the paper shows that spending is only marginally pro-poor despite the programmes aim to target only poor households. At the secondary level the programme is largely focussed on the non-poor. While most public education spending is not pro-poor it is more equitable than the overall income distribution and therefore tends to reduce income inequality. Patterns of public education spending are largely driven by differences in participation rates between quintiles; poor households tend to have lower enrolment and completion rates at all levels of the education system and this is particularly true for poor boys. If the equity of public spending is to be improved a greater focus on increasing school completion rates amongst poor students is needed.