Over the past four decades, scientists have made substantial progress in understanding the evolution of sleep patterns across the Tree of Life. Remarkably, the specifics of sleep along the human lineage have been slow to emerge. This is surprising, given our unique mental and behavioral capacity and the importance of sleep for individual cognitive performance. One view is that our species' sleep architecture is in accord with patterns documented in other mammals. We promote an alternative view, that human sleep is highly derived relative to that of other primates. Based on new and existing evidence, we specifically propose that humans are more efficient in their sleep patterns than are other primates, and that human sleep is shorter, deeper, and exhibits a higher proportion of REM than expected. Thus, we propose the sleep intensity hypothesis: Early humans experienced selective pressure to fulfill sleep needs in the shortest time possible. Several factors likely served as selective pressures for more efficient sleep, including increased predation risk in terrestrial environments, threats from intergroup conflict, and benefits arising from increased social interaction. Less sleep would enable longer active periods in which to acquire and transmit new skills and knowledge, while deeper sleep may be critical for the consolidation of those skills, leading to enhanced cognitive abilities in early humans.