As education systems around the world move towards increased accountability based on performance measures, it is important to investigate the unintended effects of accountability systems. This article seeks to explore the extent to which head teachers in a large Norwegian municipality may resort to gaming the incentive system to boost their schools' measured performances. The question addressed in this article is whether the school's selection of examinees is in line with what relevant regulations prescribe (i.e. random selection), or whether it is subject to a cream skimming of pupils (i.e. selection of those examinees who would be high achievers) and thus inducing inflation in the school's average results. The results of this study show some indications of gaming the system, in terms of the sampling of pupils in mathematics, but also indications of obeying the rules regarding the sampling of pupils in other subject matters. This article discusses implications and directions for further research. (Contains 6 notes and 5 tables.)