Spatial Augmented Reality (SAR) opens interesting perspectives for new generations of mixed reality applications. Compared to traditional human-computer interaction contexts, there is little work that studies user performance in SAR. In this paper, we present an experiment that compares pointing in SAR versus pointing in front of a screen, from standard pointing devices (mouse and graphics tablet). The results showed that the participants tend to interact in SAR in a way that is similar to the screen condition, without a big loss of performance.