This paper studies the effect of rolling out mobile money agents in rural Northern Uganda. In a randomized experiment, 168 areas were randomly selected to receive an agent in 2017, with another 163 areas serving as a control group. Administrative data on mobile money transactions suggest that the agent rollout increased the probability of sending and receiving peer-to-peer transfers. Data from a 2018 survey of more than 4,500 households show that the agent rollout led to cost-savings for remittance transactions. It also doubled the nonfarm self-employment rate, from 3.4 to 6.4 percent, and reduced the fraction of households with very low food security from 62.9 to 47.2 percent, in areas far from a bank branch. The analysis finds no effect on savings, agricultural outcomes, or poverty. Overall, the findings add new evidence that mobile money can improve livelihoods even in poor and remote settings.