This paper takes a fresh look at the issue of foreign aid fungibility. Unlike the bulk of existing empirical studies, I employ panel data that contain information on the specific purpose for which aid is given. This allows me to link aid given for education and health purposes to recipient public spending in these sectors. In addition, I attempt to distinguish between aid flows that are recorded on the recipient’s budget and those that are off-budget, and illustrate how a failure to differentiate between on- and off-budget aid produces biased estimates of fungibility. Sector programme aid is the measure of on-budget aid, while technical cooperation serves as a proxy for off-budget aid. In both sectors, across a range of specifications, technical cooperation leads to at most a small displacement of recipient public expenditure, implying limited fungibility for this type of aid. In static fixed effects models sector programme aid shows an almost one-for-one correlation with recipient public expenditure, again suggesting low fungibility, but this effect becomes imprecise and volatile in dynamic models estimated with system GMM.