With the changing political economy of social welfare, evaluation has become prevalent in the personal social services and voluntary sector organisations. This text argues that rational-technical and pluralist models of evaluation may collude with new managerialism to act as powerful processes of control. Alternative critical models of evaluation, which take account of power, are explored, so as to enable practitioners to take responsibility for evaluating practice, both in order to inhibit poor, or even corrupt, practice, and to promote good practice.