Background: Textbooks define the classical sleep cycle as an episode of non-rapid eye movement (non-REM) sleep followed by an episode of REM sleep; sometimes, a REM episode can be "skipped". While sleep cycles are considered fundamental components of sleep, their functional significance remains to a large extent unclear. One of the reasons for a lack of research progress in the field is the absence of a data-driven definition. Here, we propose to reset the scientific definition of sleep cycles on fractal (aperiodic) neural activity as a well-established marker of arousal and sleep stages, arguing that this will considerably advance the field. Methods: We used electroencephalography to compute fractal slopes and explore their temporal dynamics over the course of nocturnal sleep. We defined the "fractal cycle of sleep" as a time interval during which fractal slopes descend from their local maximum to their local minimum and then lead back to the next local maximum. Next, we assessed the correspondence between the "fractal" and "classical" sleep cycles, including "skipped" cycles. Finally, we explored fractal cycles in childhood and adolescence, a life period with ongoing sleep architecture changes, as well as in major depressive disorder, a clinical condition characterized by disturbed sleep architecture. Results: Timings of "fractal" and "classical" cycles coincided in 763/940 (81%) cases and their durations (89±34 min vs 90±25 min) correlated positively (r=0.5, p<0.001). The fractal cycle algorithm detected "skipped" cycles in 53/55 (96%) cases. In adults (range: 18-75 years, n=205), the "fractal" cycle duration and participant's age correlated negatively (r=-0.2, p=0.006). Children and adolescents (range: 8-17 years, n=21) had shorter "fractal" cycles compared to young adults (range: 23-25 years, n=24) (mean: 76±34 vs 94±32 min, p<0.001). 38 unmedicated patients with depression showed shorter "fractal" cycles compared to their own medicated state (92±38 min vs 107±51 min, p<0.001). 111 medicated patients showed longer "fractal" cycles compared to 109 matched controls (104±49 vs 88±31 min, p<0.001). Conclusions: We show that "fractal cycles" are an objective, quantifiable and biologically plausible way to display sleep neural activity and its cycles, able to provide additional information compared to hypnograms. Likewise, "fractal cycles" can be used to study the effect of antidepressants on sleep.