Differential labour market returns to male and female education are one potential explanation for large gender gaps in education in Pakistan. We empirically test this explanation by estimating private returns to education separately for male and female wage earners. This paper contributes to the literature by using a variety of methodologies (Ordinary Least Squares, Heckman correction, 2SLS and household fixed effects) in order to consistently estimate economic returns to education. The latest nationally representative data ñ the Pakistan Integrated Household Survey (2002) ñ is used. Earnings function estimates consistently reveal a sizeable gender asymmetry in economic returns to education, with returns to womenís education being substantially and statistically significantly higher than menís. The return to an additional year of schooling ranges between 7 and 11 per cent for men and between 13 and 18 per cent for women. There are also large, direct returns to womenís education at low levels of schooling and the education-earnings profile is more convex for women than men. However, a decomposition of the gender wage gap (into the component ëexplainedí by differing male and female endowments and the residual component) suggests that there is highly differentiated treatment by employers. We conclude that the total labour market returns are much higher for men, despite returns to education being higher for women. This suggests that parents may have an investment motive in allocating more resources to boys than to girls within households.