We investigate the determinants of students' university choice in Pakistan, with a focus on monetary returns, non-pecuniary factors enjoyed at school, and financial constraints.To mitigate the identification problem concerning the separation of preferences, expectations and market constraints, we use rich data on subjective expectations, with direct measures of financial constraints, to estimate a life-cycle model of school choice jointly with school-specific expectations of dropping out. We find that labor market prospects play a small role. Instead, non-pecuniary outcomes, such as the school's ideology, are the major determinants. Policy simulations suggest that relaxing financial constraints would have large welfare gains. JEL Codes: D81; D84; I21; I23. and from participants at various conferences and seminars. We are enormously indebted to our local field teams and participating institutions for without their assistance this project would not have reached its conclusion. Funding for data collection through a RAND Independent Research and Development grant is gratefully acknowledged. Delavande acknowledges funding from the Economic and Social Research Council Research Centre on Micro-social Change. d Regime with admission requirements, where a student is "admitted" only if expected graduation rank is in the top half. e Regime where the ideology of each school is made homogeneous (and is set to the IU ideology). f The model-based predicted probability of choosing each school (averaged across respondent). g Statistics in this panel are computed by assigning the student to the school with the highest predicted probability in that regime. h The median age 30 earnings change equivalent that would result in the change in utility, as a result of the change in school choice relative to the baseline case. i The equivalent change in parents' approval (on a 0-100 scale) that would result in the change in utility, as a result of the change in school choice relative to the baseline case. Mean [Median] reported. j Proportion of students who experience a utility loss under the counterfactual policy, relative to the baseline. k Proportion of students who switch to a different school under the counterfactual policy, relative to the baseline.