We measured the farm-level impacts of Malawi’s Farm Input Subsidy Programme (FISP) on fertiliser use and maize yields in central and southern Malawi. Using multiple rounds of panel data and an instrumental variable regression strategy to control for endogenous selection into the subsidy programme, we found positive and statistically significant correlations between participation in the FISP and fertiliser-use intensity. The results are broadly robust to the inclusion of previous fertiliser intensity to control for household-specific differences in fertiliser use. We combined these results with those from a maize production function to calculate programme-generated changes in average maize availability, accounting for estimated subsidy-induced changes in crop area. Our findings have implications for the way input subsidy programmes are designed and implemented.