A computer (PDP-10) simulation model was constructed using rapid, simultaneous measurements of effective refractory period (ERP), ERP dispersion (RPD), premature ventricular beat (PVB) thresholds, and multi-directional conduction times during coronary artery ligations and release in the anesthetized dog. In addition, estimates of currents of injury between ischemic and non-ischemic electrodes were included based on published data from electromagnetic recordings in dogs. Propagated PVB's were inscribed by the model when criteria for excitation, dispersion, and conduction were met based on known electrophysiological characteristics of heart muscle. The model correctly predicts high vulnerability to arrhythmias at three to seven minutes of ligation, stabilization at 10 to 15 minutes of ligation, and decreased vulnerability by lidocaine during ischemia. There was no arrhythmia when ischemic thresholds were increased by the drug before significant RPD and conduction prolongation developed. Vulnerability to arrhythmias was also predicted by the model after release of short (five minute) and long (15 minute) ligation. Since (experimentally) arrhythmias occurred much more frequently after long ligations, additional yet unknown factors other than those considered in the model must be operative in the genesis of reperfusion arrhythmias. This conclusion is supported by the observations that high ischemic thresholds induced by lidocaine returned to normal slowly after ligation release, and despite this protective effect, experimentally, lidocaine failed to abolish reperfusion arrhythmias.