Rather than a rational, technical activity, evaluation reflects the socio-political dynamics of the evaluative context. This presents a challenge for universities and the individuals within them, who may assume that plans or policies for evaluation will result in straightforward outcomes. This small-scale study in one institution looks at the tensions between institutional evaluative needs and individual evaluative practices. The results indicate that for staff in the institution, evaluative activity is largely autonomous and self-driven, rather than following institutional policy. A discretionary framework for the evaluation of learning and teaching was developed, which may be a useful tool for educational developers in their analysis of evaluative practice. (Contains 3 figures.)