In a previous issue of this journal (Vol. 3, no. 2) David Farrington introduced—‘in the interests of stimulating discussions’—an outline research design for Evaluating a Community Crime Prevention Program. This article takes up the challenge, offers a critique of his ‘quasi-experimental’ approach and insists that such programs are better evaluated using designs based on ‘scientific realist’ principles. The paper also draws attention to benefits of a ‘theories of change’ approach to evaluation, pioneered in the US in conjunction with such comprehensive community initiatives.