Conditional Cash Transfer programs are "...the world's favorite new anti-poverty device," (The Economist, July 29 2010) yet little is known about the specific role of the conditions in driving their success. In this paper, we evaluate a unique cash transfer experiment targeted at adolescent girls in Malawi that featured both a conditional (CCT) and an unconditional (UCT) treatment arm. We find that while there was a modest improvement in school enrollment in the UCT arm in comparison to the control group, this increase is only 43 percent as large as the CCT arm. The CCT arm also outperformed the UCT arm in tests of English reading comprehension. The schooling condition, however, proved costly for important non-schooling outcomes: teenage pregnancy and marriage rates were substantially higher in the CCT than the UCT arm. Our findings suggest that a CCT program for early adolescents that transitions into a UCT for older teenagers would minimize this trade-off by improving schooling outcomes while avoiding the adverse impacts of conditionality on teenage pregnancy and marriage.