A nonlinear Poincare plot method for heart rate variability (HRV) analysis is widely used to evaluate autonomic nervous system activity. However, during sleep, the Poincare plot parameters are not described completely because an ectopic heartbeat, which is similar to an artifact, depends largely on the result of the Poincare plot. Therefore, in this study, we investigated the characteristics of four Poincare plot parameters to differentiate between rapid eye movement (REM) and nonrapid eye movement (NREM) after the removal of any artifacts. For the sleep Poincare plot analysis, two 10 min target HRV segments were obtained for each of the REM and NREM sleep stages from 16 individual subjects without autonomic disease. With a 2 min moving window with a time shift of 2 s, 240 results were created and led to the formation of cloud points (or ellipses) on the Cartesian coordinate system. The length (SD1) and width (SD2) of an ellipse, ellipse area (Ln sArea), and SD2/SD1 ratio were calculated to obtain the mean value and standard deviation (SD) of all the results. The SD2/SD1 ratio and Ln sArea exhibited excellent discrimination ability between REM and NREM in terms of their mean and SD (p = 0.0001 and p = 0.0003, respectively; 95% CI), but SD1 did not (p = 0.9606 for mean and p = 0.0320 for SD). The Ln sArea remained nearly constant and was independent of the sleep stages. The results indicate that the Ln sArea and SD2/SD1 ratio could be used to assess physiological changes and health status during sleep and to discriminate between REM and NREM, respectively.