McCarley and Massaquoi successfully simulated human REM-NREM cycle characteristics by extending the McCarley-Hobson model with two sets of assumptions, one creating limit cycle behavior, the other introducing two sources of circadian variation. We argue that the limit cycle assumptions, due to freedom in choosing parameter values, suffice to explain variation in REM across the night. Nonmonotonic circadian variation in REM latency requires a circadian cycle dependence only of initial conditions at sleep onset.