Considering the stochastic nature of human circadian rhythms, it is reasonable to consider the use of Box‐Jenkins time series approach, for modeling these rhythms. This approach is based on fitting an auto regressive moving average model to the data and using the model to forecast future values. The general approach is described in detail and is then applied to human rectal temperature circadian rhythms to show its usefulness and versatility in chronobiological modeling.