Objective: The objective of the paper is to verify the accuracy of simulated motion from key frames and to find the way to reduce the storage taken by motion data while maintaining the accuracy of digital human modeling simulation (DHMS). Background: Digital human modeling (DHM) tools have been widely used to ergonomically design and evaluate various human-involved systems (e.g., vehicles). DHM tools can also be used for simulating drivers’ ingress and egress motions, enabling vehicle engineers to design and evaluate vehicle configuration around the car door. Advanced motion capture technology has contributed to obtaining and reproducing realistic human motion data. Due to high frame rate typically required for accurately capturing dynamic human motion (e.g., ingress/egress), however, it often takes too much time and memory space for DHMS systems to deal with motion data. In order to save calculation time and memory space, a motion simulation method based on key-frames is proposed in this study. Method: Accurate prediction of foot trajectories is critical when predicting accurate driver ingress/egress motion. This paper presents an approach to improve the accuracy of foot trajectory simulation based on key frames of driver motion data obtained from a laboratory study. Twenty-one people completed the study, and their motions were recorded while they entered/exited a reconfigurable vehicle mockup. Motion capture data from optical type motion capture system (Hawk (i) ) were compared to simulated data processed by RAMSIS™ digital human modeling software. Results: As to the full-frame ingress and egress motion, the difference between raw data and simulated motion data was not statistically significant (p≥ 0.807). However, the difference between raw data and simulation data tend to increase as the movement speed increases. One-way ANOVA was conducted to see the differences between raw data and simulated data at the key-frame moments. The results show that there are significant differences between the two different data sources for both ingress and egress situations (p < 0.000). The author suggests regression estimations to improve the accuracy of the output from simulation. Conclusion: Although the differences between raw data and simulated motion data were not significant, it requires refinement especially for the key-frames. Application: The results can be applied to preprocessing or postprocessing of motion capture data when using RAMSIS™.