Two frames of reference (exocentric and egocentric) and two points of view (first-person and second-person) were contrasted in two experiments to measure their effect on wayfinding performance. Twelve male infantry soldiers used three helmet-mounted display designs (second-person exocentric, second-person egocentric, first-person egocentric) to locate bearings to waypoints. From a constant location, participants were required to face waypoints presented at offset bearings of 25°, 65° and 125°. The egocentric display supported faster performance times than the exocentric display, significantly faster at the largest offset. Faster times were attributed to fewer head misdirection errors and better display compatibility with expectations of spatial orientation. The second-person display supported significantly faster performance times than the first-person display at all offset levels. However, the time differences were not attributed to accuracy or head misdirection errors. Other display features, such as amount of information or field of view may have contributed to the differences. There were no qualitative differences between displays for ease of use, acceptance for land navigation, or mental demand. Implications for the design of helmet-mounted displays for land-based navigational tasks are discussed.