A mutual information (MI) method for assessment of the relationship between hemodynamic variables was proposed and applied to the analysis of heart rate (HR), arterial blood pressure (BP), and renal sympathetic nerve activity (RSNA) in artificial heart-implanted dogs to quantify correlation between these parameters. MI measures the nonlinear as well as linear dependence of two variables. Simulation studies revealed that this MI technique furnishes mathematical features well suited to the investigation of nonlinear dynamics such as the cardiovascular system and can quantify a relationship between two parameters. To constitute a model free of the natural heart, two pneumatically actuated ventricular assist devices were implanted as biventricular bypasses in acute canine experiments. RSNA was detected with the use of bipolar electrodes attached to the renal sympathetic nerve. Analysis of data during control revealed that correlation between HR and RSNA was higher than that between HR and BP and that between RSNA and BP (P < 0.05). Although RSNA seemed to fluctuate noncorrelatedly with BP in higher pacing rates, the MI values between them disclosed their strong correlation. Surprisingly, correlation between RSNA and BP was stronger during a pacing rate of 60 beats/min than during higher pacing rates and control (P < 0. 05). It is suggested that the baroreflex system may be susceptible to pacing rates during the total artificial heart state. We calculated the time delay between HR and RSNA, between RSNA and BP, and between HR and BP by regarding a time delay at which the maximum MI value between each pair of parameters was given as a physiological delay. Our results indicate that RSNA leads BP, BP leads HR, and RSNA leads HR during control (P < 0.05). We conclude that this method could provide a powerful means for measuring correlation of physiological variables.