Fertiliser intended for government subsidy programmes is sometimes diverted and sold to farmers at or near market prices. Failure to account for such ‘leakage’ can upwardly bias econometric estimates of the effect of government fertiliser subsidy programmes on total fertiliser use. This paper extends the framework used in earlier studies on the crowding in/crowding out effects of subsidised fertiliser on commercial fertiliser purchases to account for leakage, and then applies it to the case of Zambia. Results suggest that each additional kg of subsidised fertiliser injected into the system increases total fertiliser use by 0.54 kg. Without controlling for leakage, the estimate would have been 0.87, an overestimate of 61%.