The authors' findings show that quantitative measures of complexity (correlation dimension, CD) and predictability (Lyapunov exponent, LE) provide significant information about autonomic nervous system (ANS) processes. There is well-organized nonlinear behavior of heart rate variability (HRV) and respiratory movements (RESP), which can be interpreted with regard to terms such as nonlinear stochastic, regular deterministic, and chaotic. The clear identification of a physiological process only on the basis of a measured time series is difficult. Distinguishing between chaotic and nonlinear correlated stochastic processes, in particular, needs more information than that of a positive LE, noninteger CD, and nonlinearity. The additional considerations of model investigation and phase locking make chaotic underlying processes of HRV and RESP in S1 probable. The hypothesis that deviations from the normal function lead to a decreased complexity and increased predictability could be confirmed quantitatively by the estimation of CD and LE during S2 and S3. This information, which can not be found by a linear approach to time series analysis, is important for the understanding of normal and pathologically disturbed functions. The authors do not claim that their analysis replaces linear methods, but rather that a consideration of both linear and nonlinear properties may improve diagnostic classifications. The potential usefulness of dynamic nonlinear analysis presented herein is in the improved understanding of the complex processes of the ANS, and in the resulting medical concepts with regard to pathophysiological disturbances and their treatment.