The purpose of this study was to investigate the effectiveness of the Problem Based Learning (PBL) supported by interactive computer simulations within the context of "Buoyant Force in liquids and gases" subject on academic performance. The study was conducted with a quantitative methodology via a type of quasi-experimental design. The sample of 54 students enrolled in two eighth grade classes of a public middle school in Turkey. One of these classes was randomly assigned as experimental group and instructed by means of the PBL, whereas the other class has been assigned as control group and instructed by means of traditionally designed instruction. In order to measure academic performance of the students, Academic Performance Test was used as pre- and post-tests in both control and experimental groups. The results of MANOVA showed that the students in PBL classes had significantly higher mean scores on academic achievement whereas there was no difference between the groups on performance skills.