This experimental study was intended to examine whether game-based learning (GBL) that encompasses four particular game characteristics (challenges, a storyline, immediate rewards and the integration of game-play with learning content) in an OpenSimulator-supported virtual reality learning environment can improve perceived motivational quality of the learning activity (P-MQLA) for elementary school students. In this pre- and post-test experimental comparison study, data were collected from 132 fourth-graders through a Short Instructional Materials Motivation Survey. The same tasks were provided for the experimental and control groups, except that the control group's tasks did not include the four aforementioned gaming characteristics. The non-GBL for the control group involved mainly web-based word problems such as selecting the correct answer by clicking on the potential responses. P-MQLA was improved after the intervention in both groups. There is a trend that the experimental group improved the P-MQLA more than the control group, at borderline statistical significance. The analysis of covariance also indicated a significant interaction between the group variable and the pre-intervention P-MQLA, indicating that the effect of the virtual reality-supported GBL is moderated by learners' pre-intervention report of P-MQLA.