Virtual reality-based instruction such as virtual worlds, games, and simulations are becoming very popular in K-12 and higher education. Three manuscripts that report the results of investigations of these increasingly prevalent instructional media were developed for this dissertation. The purpose of the first study, a meta-analysis, was to analyze the instructional effectiveness of virtual reality-based instruction when compared to the traditional methods of instruction. In addition, this study also explored selected instructional design features of the virtual learning environment that moderated the relationship between instructional method and the academic achievements. Analyses of 63 experimental or quasi-experimental studies that studied learning outcomes of virtual reality-based instruction in K-12 or higher education settings yielded a mean effect size of g = 0.47 (SE = 0.02) suggesting that virtual reality-based instruction is an effective medium of delivering instruction. Further analyses examined factors that influence its effectiveness. The purpose of the second study was to examine a model of the impact of a 3-D desktop virtual reality environment on the learner characteristics (i.e. perceptual and psychological variables) that can enhance chemistry-related learning achievements in an introductory college chemistry class. A theoretical model of the relationships of features of 3-D virtual reality environments and students' experiences in the environments to outcomes on a chemistry learning test and measures of spatial ability and self-efficacy was tested using structural equation modeling. Usability strongly mediated the relationship between 3-D virtual reality features, spatial orientation, self-efficacy, and presence. Spatial orientation and self-efficacy had a statistically significant, positive impact on the chemistry learning test. The purpose of the third study was to investigate the potential of Second Life (SL), a 3-D virtual world, to enhance undergraduate students? learning of a foundational chemistry concept, spatial ability, and self-efficacy. A quasi-experimental pretest-posttest control group design was used. A total of 387 participants completed three assignment activities either in Second Life or using 2-D images. The difference between the scores of 3-D virtual environment-based group and the 2-D images-based group was not statistically significant for any of the measures.