To study heart rate baroreflex gain, autoregressive moving average (ARMA) analysis, a multivariate method that allows evaluation of the dynamic ("beat-to-beat") interactions between changes in biological signals, was used to evaluate the relationships between R-R interval and arterial blood pressure (BP) during random-interval breathing. Parameters obtained by ARMA analysis of spontaneous fluctuations in BP and R-R interval in 17 volunteers were used to model the response of R-R interval to a transient 1-mmHg increase in BP; the resulting impulse-response and step-response curves were compared with baroreflex gain measured using bolus injections of phenylephrine (PE) and sodium nitroprusside (SNP). Impulse-response curves for the systolic BP-R-R relationship showed an early (0-1 s) sharp maximum of 5.5 +/- 4.2 ms/mmHg, which was smaller in magnitude but linearly correlated with baroreflex gain derived from SNP (14.5 +/- 9.7 ms/mmHg; r = 0.80, P < 0.002) and PE (31.6 +/- 26.7 ms/mmHg; r = 0.53, P < 0.05) injections. A similar relationship was also found between the one-beat ARMA step response and SNP injection (r = 0.70, P = 0.01). The integrated step response of the BP-R-R relationship over 6 s was 6.4 +/- 4.1 ms/mmHg, with no correlation to baroreflex gain determined by SNP (r = 0.33, P = 0.20) or PE (r = -0.15, P = 0.57). In conclusion, quantification of baroreflex gain consistent with other techniques may be achieved by ARMA analysis without perturbing mean BP. Correlation of baroreflex gain obtained by bolus injection to early measures of baroreflex gain obtained from the ARMA maximum impulse and early step responses, but not the late step response, suggests that the ARMA method may provide additional information regarding the frequency dependent effects of BP on R-R-interval.