Recent technological advances allow symbology to be displayed on the pilot's visor. A major benefit of this is that the pilot will be able to take this information with them when they look off-boresight. However, when looking off-boresight the question arises as to what is the best orientation, or frame of reference, for attitude symbology against the horizon (i.e., forward or line-of-sight) in order to maximize interpretation and performance. This study tested five different symbologies (standard HUD, visually coupled acquisition and targeting symbology, arc segmented attitude reference, theta ball, and non-distributed flight reference) of which three have both forward and line-of-sight orientations. The experiment consisted of two different tasks, with the pilots performing either facing the monitor or rotated 90 degree(s) and looking over their shoulder (off-boresight). In the first task, pilots maintained straight and level flight with simulated turbulence. The second task had pilots interpret a static representation of their attitude and respond via a key press, and then the display went live and they had to fly to a new commanded attitude. This second task was similar to a recovery from unusual attitude methodology, except the end state was never straight and level. Instead, a second unknown end state attitude was commanded by the experiment. Results indicate that performance is better when the symbology is forward as opposed to line-of-sight referenced. Further, performance was best in both tasks for the non-distributed flight reference. We discuss these results in terms of implications for helmet-mounted display symbology design.