Extensive research has been done on external human-machine interfaces, and a growing number of studies have been conducted measuring eye tracking. This study investigated different eye-tracking measures and their relation to pedestrian crossing initiation time in virtual reality. The objective is to understand further the applications and differences between measuring saccade and fixation for examining pedestrian crossing behavior. The pedestrian crossing initiation time, saccade count, fixation length, and fixation count were examined with the HTC Pro Eye virtual reality eye-tracking headset (N = 30). Significant differences were found between eHMI type, crossing initiation time, and all eye metrics (fixation length, fixation count, and saccade count). Interaction effects were found between crossing initiation time and saccade count. Saccade count was found to be the most appropriate eye-tracking measure for pedestrian crossing studies, followed by fixation length and then fixation count. Furthermore, it is not necessary to include both saccade and fixation. This study provides recommendations for future eye-tracking research, especially in human-vehicle interaction.