Changes in the sequence or pattern of R-R intervals is an integral part of electrocardiogram detection of most clinically important arrhythmias. Moreover, there is inherent interest in these patterns and their method of study as basic physiological phenomena. Prediction of R-R interval changes based on past history and detection of departure from an established pattern have been utilized in automated monitoring strategies and algorithms. The studies described her further clarify the usefulness of Markov processes for this purpose, compare the accuracy of linear and nonlinear regression models, and indicate some advantages of both.