Objective: A radial arm maze (RAM) is an essential tool for assessing spatial learning and memory. Although this tool is widely used to study deficits in spatial memory in animal models, it has several restrictions that prevent its adaptation to human research and training. Therefore, we developed a head-mounted-display RAM (HMD-RAM) program for humans and verified its validity by comparing it to the results obtained by previous RAM studies. We also compared the HMD and a flat monitor as experimental devices. Methods: Forty participants were recruited for the current study (Study 1: 20 participants with the HMD device; Study 2: 20 participants with the flat monitor). They navigated a virtual room as a first-person viewer and used environmental landmarks to remember their spatial position and orientation. The main dependent measures were working memory error, reference memory error, detection time, travel distance, and participant’s head movements. To validate the program, participants also conducted neuropsychological assessments and self-reported measures. Results: The results for HMD-RAM tasks were consistent with the results of previous research conducted on animals, and the HMD elicited a higher sense of presence, immersion, and simulator sickness than the flat monitor. According to post-experiment questions on navigation strategy, creating landmarks was important when people were discovering locations in their environment, and an HMD was beneficial for better navigation strategy. Conclusion: These results suggest that the HMD-RAM is valuable for estimating spatial learning and memory in humans and may be a useful tool for early diagnosis of deficits in spatial learning and memory, including amnestic mild cognitive impairment and Alzheimer’s disease.