An experimental study involving 30 undergraduates (mean age = 20.5 years) in mental rotation (MR) training was conducted in an interactive Desktop Mental Rotation Trainer (iDeMRT). Stratified random sampling assigned students into one experimental group and one control group. The former trained in iDeMRT and the latter trained in conventional condition. A multifactorial pretest posttest design procedure was used and data were analysed using two-way analysis of covariance. Overall, there was substantial improvement in MR accuracy. Main effects of training and gender were observed, indicating that iDeMRT group and boys outperformed the control group and girls respectively. In addition, an interaction between training method and gender was present, indicating that boys were more accurate when trained in iDeMRT and performed moderately in conventional method. Female participants achieved equivalent improvement gain in MR accuracy regardless of the training conditions used. For the speed measure of MR, no appreciable improvement was observed after training.