Résumé : Study Objectives: We investigated the effects of the variation in the number of NREMS episodes on human sleep, in particular on delta activity distribution. Participants: Forty-one medication-free healthy men without personal or family history of psychiatric disorders. Interventions: Subjects slept four consecutive nights, the last three were analyzed. Sleep was undisturbed and uninterrupted following self-selected schedule. Results: With the exception of one night out of 123, subjects slept at least four NREMS episodes with 80 % presenting five or six episodes per night. Yet, only 24 % slept the same number of NREMS episodes across three nights. Shorter first NREMS episode was associated with greater number of NREMS episodes, and the total number of NREMS episodes was significantly predicted by the duration of the first one. Whether subjects slept four, five or six NREMS episodes, the proportion of TST spent in REMS, total delta power, total spectral power and the rate of delta power in the first NREMS episode did not differ between subjects. In each night, the distribution of delta activity across the first four NREMS episodes was not modified by the total number of NREMS episodes. Conclusion: When spontaneous sleep is allowed, healthy men mostly sleep beyond the fourth NREMS episode. Based on delta activity distribution across the first four episodes, we have demonstrated that the number of NREMS episodes does not modify delta sleep homeostasis. Only duration of first NREMS episode predicts the number of NREMS episodes slept. These results suggest that sleep homeostasis and NREMS-REMS alternation may be two independent processes.