In SAE levels 4 and 5 autonomous vehicles (AVs), drivers are no longer required to control the vehicle, which may limit communication with pedestrians. To address this, external human-machine interfaces (eHMI) have been introduced as a means of facilitating interaction between AVs and pedestrians. The advisory and directive types of eHMI messages have been reported as effective in clearly conveying specific actions to pedestrians. However, these messages pose legal and ethical challenges and may result in various risks depending on the surrounding context. Therefore, this study analyzed the impact of eHMI messages conveying the movement intention of AVs on pedestrians’ crossing decisions. A virtual reality experiment was conducted with 40 participants to evaluate pedestrians’ crossing experiences based on eHMI message types and AV braking styles. Results indicated a tendency for movement information-based eHMI messages to reduce the pedestrian crossing decision time (CDT), although the difference was not statistically significant. However, a significant difference was observed in the CDT depending on the AV’s braking style. These findings suggested that although eHMI messages may partially contribute to pedestrians’ crossing decisions, the primary determinant remains the physical movement of the AV.