The Government of Nicaragua and the US foreign assistance agency, the Millennium Challenge Corporation, launched a rural business services program designed to boost the income of the small farm sector. Relying on a randomized rollout strategy, this paper reports the results of a multi-year impact evaluation that spanned the 5-year life of the program. We argue that impacts of a program of this sort are unlikely to be fully revealed by standard binary treatment estimators and show that temporal pattern of impact indeed evolves in important ways over time. Income in the activities targeted by the program steadily rose, plateauing at a 30% increase over baseline after two years in the program. The program also appears to have provoked significant increases in both mobile and perhaps fixed farm capital. However, on average there have been no significant impacts on household living standards to date. Finally, conditional quantile regressions give evidence of quite substantial heterogeneity in program impact.