Use ofa long-run profit function provides accurate estimates of returns to agricultural R&D in Europe and North America, a result that is tied to the long term basic and applied nature of the research institutions in these areas. Agricultural R&D in South Africa is fundamentally different with a more adaptive character, resulting in a shorter lag of five years between research spending and changes in TFP. In this situation, the long-run profit function input demand and output supply elasticities are too high, resulting in overestimation of returns to R&D. A potential solution to this problem is to use a short-run profit function in the calculation of returns to R&D. Several other inputs are held fixed and the result is a short-run return to R&D in South Africa that compares favourably with long-run return levels that have been established for the northern hemisphere.