The subject of evaluation is important to research managers generally and especially to those concerned with the formulation and direction of programmes. Evaluation is normally impracticable at the programme level; instead attention must be concentrated at the level of the project. The purposes served by such evaluation are three: to increase the awareness of research managers about the likely impacts of their choices, to provide a more critical basis for varying or stopping projects and to improve the choice of new projects and their design. The lack of appropriate data limits the methods available for evaluating forestry research to cost-benefit analysis. The components of total benefits distinguished are: expected contributions to outputs which can be priced, a score for expected benefits in the field of environmental outputs which cannot currently be priced, and a score for scientific value which is defined as the contribution to knowledge not captured in the other two components of total benefits. While these three elements can stand alone, their combination in a single measure is useful and methods are described for calculation of weights to make them commensurable. The results of assessing 45 current projects in the Research Division’s programme are presented together with observations on the effect of differing weights on project ranking. Needs for further work identified are: more applications of available techniques for valuing environmental benefits in such fields as wildlife conservation, practice in peer review and the estimation of scientific merit, and experience in the assessment of the probability of success of achieving project objectives.