Human movement has been found to have positive effects on learning performance. This study examined the effects of using Dynamic Geometry Software (DGS) CABRI to manipulate geometric properties of triangles or observing those manipulations made by an instructor on learning geometric properties with DGS-CABRI. Participants were 60 year 5 students, who received instructions on geometric problems and were randomly assigned to three conditions: A condition in which they performed mouse movements to manipulate geometric properties of triangles, a condition in which they observed the teacher performing those manipulations, and a conventional condition in which they studied a static format of the learning materials without any manipulations. We hypothesized that learning conditions involving manipulations of geometric properties of triangles would result in lower cognitive load and higher performance on a retention and transfer test than the conventional condition. Moreover, we hypothesized that making manipulations of the geometric properties of triangles through mouse movements would be superior to observing those manipulations being made by an instructor in terms of cognitive load, retention- and transfer test performance. Whereas the first hypothesis was confirmed, the latter hypothesis was only confirmed for retention test performance. Possible implications for educational practice are discussed.