Sleep, a critical necessity for health, cognition, and well-being, consumes 25 years of the average human’s life. Alterations in sleep’s macro and microstate have been observed in multiple illnesses, including depression and Alzheimer’s disease, and sleep may represent a biomarker of these difficult to diagnose pathologies. Despite this, adequate models of the dynamics of sleep able to capture the fine-grained minutia of sleep’s many states and processes are lacking. The current work leverages over 1000 nights of healthy sleep polysomnography recordings from multiple open-source datasets. Data were automatically analyzed through a machine learning and data engineering pipeline to extract important features of the electroencephalogram, such as sleep spindles, slow oscillations (SO), and Rapid Eye Movement (REM) events. The counts of these sleep features and transition probabilities of sleep stages were modeled across the night in 30-second epoch intervals, using a flexible yet principled Bayesian cognitive modeling approach. The model, christened P(Sleep) 2.0, defines “typical” patterns of healthy sleep, and how they are affected by time, circadian processes and individual differences (e.g., age, sex). Transition probability dynamics were found to best be predicted by the previous 2 stages, time spent in stage, time across the night, age and sex, but not an age*sex interaction. Spindles, SO and REM events were also affected by the same features, but an age*sex interaction was additionally present. N2 spindles and both N2 and N3 slow oscillations were found to increase with time spent in a stage, whereas N3 spindles and REM decreased. Spindles and SO were more present in the evening compared to the morning; in contrast, REM increased over the night. REM counts and N3 spindles increased with age, while N2 spindles and SO reduced. Females had more N2 spindles, and SO in both N2 and N3 than men. The current work replicates a body of previous literature based on small sample sizes, and the increased power in this study provides firm conclusions to arguments on the patterns of healthy sleep across the night, both at the macroscale (stages) and microscale (feature events). The proposed work has broader impacts on the detection of abnormal (and potentially pathologic) sleep patterns.