Early marriage is a concern for many developing countries. As an incentive to increase school enrolment and delay age for marriage many countries have implemented conditional cash transfer programs for girls in secondary school. The Female Secondary Stipend Program (FSSP) in Bangladesh introduced in 1994 was one of the pioneer such programs. There are many studies on the effects of stipend programs on school enrolment, labor market participation and poverty reduction but not on causal effects on age at marriage. This paper studies the effects of FSSP as an exogenous variation in time and region on both schooling outcomes and age at first marriage with regression discontinuity and difference-in-difference approaches. The regression discontinuity results show that the program increased completed years of schooling by at least 0.4 year and age at first marriage is delayed by at least 0.4 year. We also show that the difference in difference method predicts biased result.