This paper evaluates the effects of microcredit on household consumption using a large dataset from Bangladesh. Village fixed effects and instrumental variable strategies are used to estimate the causal effects of microcredit program participation. Overall, the results indicate that the effects of microcredit on consumption vary across different groups of poor household borrowers. The groups that benefit the most include the poorest of the poor participants. The benefits are low for households that are marginal to the participation decision. The effects of participation are generally stronger for female borrowers than for male borrowers.