Most studies concerned with measuring the rate of return to publicly-funded agricultural R&D investment have found high returns, suggesting under-investment, and calls for increased expenditure have been common. However, the evaluation of returns tends to measure the effect of research expenditure against growth in total factor productivity (TFP), based on market inputs and outputs. When compared against growing public unease over the environmental effects of pursuing agricultural productivity growth, TFP indices become a misleading measure of growth. This paper integrates some non-market components into the TFP index. The costs of two specific externalities of agricultural production, namely fertiliser and pesticide pollution, are integrated in a TFP index constructed for the period 1948-1995. This adjusted, or 'social', TFP index is measured against UK public R&D expenditures. The rates of return to agricultural R&D are reduced by using the 'social' as opposed to the traditional TFP index. Whilst both remain at justifiable levels, previous studies appear to have over-estimated the effect of agricultural R&D expenditures. Furthermore, with changes in policy towards more socially acceptable but non-productivity enhancing outcomes, such as animal welfare, rural diversification and organic farming, the future framework for analysing returns to agricultural R&D should not be so dependent on productivity growth as an indicator of research effectiveness. © 2002 Elsevier Science B.V. All rights reserved.