This study explored ways of developing and utilizing more efficient Computer-Based Instructional Simulations (CBIS) based on the relationship among fidelity levels (i.e., how closely a simulation imitates reality), transfer time, and learning time. The hypotheses presented suggest how to determine the optimal shifting points of the fidelity levels; this method yields the optimal routes that would help learners reach goals with less time in CBIS. To verify the hypotheses, 148 seventh grade male students were selected voluntarily from one junior high school in Korea, and a 4 X 6 (Fidelity Levels X Number of Lessons) posttest-only experimental design was employed. Even though the result failed to support main hypotheses due to no distinct difference among fidelity levels of CBIS, the result shows that a cost-effective point on learning time exists in a given level of fidelity range of CBIS. Therefore, the hypotheses need to be tested again in further well-designed experiments with a larger range of fidelity. Figures illustrate the relationship between transfer of learning and learning time according to fidelity levels and the optimal shifting of fidelity levels and its benefit. A table of the standards for classifying fidelity levels is included. (Contains 23 references.) (Author/DLS)