The temporal structure of human sleep changes across development as it consolidates from the polyphasic sleep of infants to the single nighttime sleep episode typical in adults. Experimental studies have shown that changes in the dynamics of sleep need may mediate this developmental transition in sleep patterning, however, it is unknown how sleep architecture interacts with these changes. We employ a physiologically-based mathematical model that generates wake, rapid eye movement (REM) and non-REM (NREM) sleep states to investigate how NREM–REM alternation affects the transition in sleep patterns as the dynamics of the homeostatic sleep drive are varied. To study the mechanisms producing these transitions, we analyze the bifurcations of numerically-computed circle maps that represent key dynamics of the full sleep–wake network model by tracking the evolution of sleep onsets across different circadian (∼ 24 h) phases. The maps are non-monotonic and discontinuous, being composed of branches that correspond to sleep–wake cycles containing distinct numbers of REM bouts. As the rates of accumulation and decay of the homeostatic sleep drive are varied, we identify the bifurcations that disrupt a period-adding-like behavior of sleep patterns in the transition between biphasic and monophasic sleep. These bifurcations include border collision and saddle–node bifurcations that initiate new sleep patterns, period-doubling bifurcations leading to higher-order patterns of NREM–REM alternation, and intervals of bistability of sleep patterns with different NREM–REM alternations. Furthermore, patterns of NREM–REM alternation exhibit variable behaviors in different regimes of constant sleep–wake patterns. Overall, the sequence of sleep–wake behaviors, and underlying bifurcations, in the transition from biphasic to monophasic sleep in this three-state model is more complex than behavior observed in models of sleep–wake regulation that do not consider the dynamics of NREM–REM alternation. These results suggest that interactions between the dynamics of the homeostatic sleep drive and the dynamics of NREM–REM alternation may contribute to the wide interindividual variation observed when young children transition from napping to non-napping behavior.