Programs to fight poverty aim at allowing individuals to support themselves ex-post, when they are not part of the program anymore. We compare the ex-post effects of conditional and unconditional cash transfer programs on labor income. We use an experiment where low-income individuals are randomly assigned to three groups: no treatment, unconditional cash transfer, and cash transfer conditional on reemployment training. We exploit Social Security data, including all registered labor contracts in Italy. Results show that conditional cash transfers have positive and sizeable effects on labor income, both contemporaneous and ex-post effects. These effects last at least two years and are led by males. Unconditional cash transfers have no impact on labor income.