ABSTRACTEmploying a semi‐Markov model, all‐night sleep‐wake patterns of 68 full‐term, normal infants, 32 at 2 mos and 36 at 9 mos of age, were analyzed. The results are compared to the changes in proportional relationships of sleep‐wake states with age and to the time series method of binary autocorrelation. Simulated nights of sleep were also computed using parameters derived from the semi‐Markov model.We conclude that the sleep process in infants meets the assumptions of a semi‐Markov model and that using such a model provides information about the maturation of transition and sequencing mechanisms of sleep‐wake states. Simulating sleep patterns offers an opportunity to model hypothetical disturbances of sleep and assess the effectiveness of potentia