With the renewed interest in VR, new questions arise for content creators, as existing cinematic practices cannot simply be transferred. In this paper, we describe two experiments investigating which voice-over perspective elicits the best sense of presence for viewers of cinematic VR content. For the first experiment different voice-over narrations in first, second and third person perspectives were added to a VR video. This test showed that viewers preferred the voice-over in second person perspective, as this provided them with the strongest sense of presence and a feeling of 'being in the story'. In the second experiment, we used a short 360° documentary with a first person voice-over perspective, and compared it to a version of the same documentary with a second person voice-over, using a quantitative survey. In this experiment, however, no significant difference was found between the two groups of respondents. In our discussion, we explore several possible reasons that may have contributed to this outcome.