In the present study, we propose a new technique for estimating depth of sleep over the whole night using electrocardiogram (ECG) RR intervals (RRIs). We produced a Lorenz plot (LP) using the RRIs recorded during all-night sleep and confirmed that changes in distribution on the LP occur based on changes in sleep stage. To evaluate the changes in these distributions, RRIs are projected a LP on a y = x axis, y = -x axis, and analyzed the shifting of the mean (center C) and standard deviation (area S) for each sleep stage. Analysis interval time was 60 seconds, shifting every second, and we compared heart rate variability (HRV) and sleep level. Center C showed progress toward light sleep levels and area S showed the transition phases toward deep sleep. A concordance rate of 60.1% between the estimated values and actual transitional sleep level was obtained for all-night sleep. Therefore, transitional sleep level can be evaluated based on HRV using the LP.