Over the last five to ten years Europe has witnessed a return to fashion for evaluation. As an activity it has been expanded in most EU states and very considerably within the EU Commission. Within this broad trend there has been a strong sub-fashion for collaborative, 'pluralist' and 'partenarial' approaches to evaluation. This article comments on the nature of these approaches, notes some of the claims made on their behalf, and identifies situations in which collaboration is likely to decrease both the scientific rigour and the usefulness of an evaluation. In certain circumstances, the 'old-fashioned' model of an independent, scientific, form of evaluation is to be preferred to models of pluralism, participation and empowerment.