This study examined whether the proportion of time spent on varying positions could account for the accuracy of self-perceived sleep position. The sample contained 26 healthy subjects, who were invited to sleep at a laboratory for two consecutive nights and in the next morning, were asked to answer some questions about their sleep, including self-perceived major sleep position and other sleep positions adopted. Their sleep was videotaped and was coded by two external judges. For both laboratory nights, subjects who correctly reported their major sleep position spent a larger proportion of their sleep time on the major body position, spent a smaller proportion of their sleep time on positions other than the major one, showed a larger discrepancy between the two proportions, and tended to rest their hands on their chest or stomach during sleep rather than put their hands aside their trunk than did subjects who incorrectly identified their major sleep position. These findings suggest that the misperception of one’s sleep position can be attributed to the individual difference in the variability of sleep positions across the night. Although most subjects could correctly identify their major sleep position, the inaccuracy rate was high – that is, 23.1% for the first laboratory night and 42.3% for the second laboratory night. Furthermore, self-perceived and externally coded sleep positions were not significantly associated in the second night. The self-report method is far less expensive than the video-coding method but researchers should be cautious about its limitations and consistency with objective measures. Since the body posture during sleep is dynamic rather than static and sleepers may not be aware of their position shifts across the night, self-reported sleep position does not necessarily correspond to its objective counterpart, especially for those sleepers whose spend a similar proportion of time lying in different positions across the night. For clinical purposes, therefore, self-report of sleep position should be complemented by video records. Some specific body postures – such as resting hands on the chest and the nose pointing to one side – were demonstrated to be differentially correlated with supine and lateral sleep. Investigators may make use of these additional body postures to improve the accuracy of a reported sleep position.