This article reports the findings of an experimental study involving 36 primary school students (16 girls, 20 boys, Mean age = 9.5 years, age range: 8-10 years) in geometrical understanding of 2D and 3D objects. Students were assigned into two experimental groups and one control group based on a stratified random sampling procedure. The first experimental group trained in an interaction-enabled trainer, the second experimental group trained in animation-enhanced trainer, and the control group trained using printed materials. A multi-factorial pretest posttest design procedure was used and data were analyzed using 2-way ANCOVA. Results indicate that there was substantial improvement of geometrical understanding after training. Effects attributed to training method were detected where the experimental groups particularly the interaction-enabled group achieved higher gains compared to the control group. Gender difference was also detected where in general boys achieved higher geometrical understanding compared to girls. Interaction effects between method of training and gender were observed where interaction condition favored boys, animation condition was gender-neutral, and conventional condition disadvantaged boys. (Contains 5 tables and 1 note.)