Three algorithms for assessment of respiratory sinus arrhythmia (RSA) have been evaluated: cross-correlation function, histogram analysis and regression plot. The algorithms were tested experimentally in a group of 11 subjects. A cross-correlation function with a high time resolution (1 ms) was used for investigation of the time lag between instantaneous heart rate and respiration (CTL). This time lag was not affected by the breathing rate in a range of 8 to 29 breaths per minute. A mathematical model of CTL compared with experimental results indicates that respiratory sinus arrhythmia is probably modulated directly by the respiratory network in the brainstem rather than by a baroreflex in the range of breathing rate investigated. Histogram analysis reflects the impact of inspiration and expiration on respiratory sinus arrhythmia. For this purpose heart rate changes were separated into two distributions (inspiration - expiration). The result value (CI-VAL) of the Mann-Whitney (J-test reflects the impact of respiration on heart rate variability. Regression analysis of heart rate versus respiration shows that the heart rate increase is more closely coupled to inspiration than the heart rate decrease to expiration. Both, CTL and CJ-VAL are thought to be useful parameters for clinical investigation of RSA.