Antiarrhythmic drug effects on maximal upstroke velocity (Vmax) are frequency dependent, which implies that the effects of these drugs on conduction should also be rate dependent. Previous in vivo studies have been limited by assumptions about unchanging propagation pathway, and by the empirical use of a first-order recovery model. To explore time-dependent antiarrhythmic drug-induced conduction slowing in vivo, we used 56-electrode epicardial mapping in chloralose-anesthetized dogs with formalin-induced atrioventricular block. Intervaldependent changes in conduction time were assessed under control conditions and then after three loading and maintenance infusions of procainamide. Under control conditions, epicardial activation time (86 ± 26 msec at a basic cycle length of 300 msec) was unchanged (87 ± 24 msec) by pauses up to 6.6 ± 2.2 seconds. Procainamide caused conduction slowing that dissipated as a function of recovery interval, with 94 ± 6% recovery over a maximum pause of 6.7 ± 1.5 seconds, but did not alter activation pattern. Drug-induced changes in conduction were evaluated by use of a mathematical model assuming phase 0 inward current proportional to conduction velocity squared. Conduction changes were better fitted by this "quadratic model" (least sum of squared deviations 3.9 ×l0−3by mapping in five dogs, 2.7 ×l0−3by use of QRS duration in nine dogs) than by a monoexponential model (sum of squared deviations 5.7×l0−3by mapping, 3.4×l0−2with QRS;<0.01 vs. quadratic model for each). As predicted by theoretical analysis, recovery time constants from the quadratic model were similar to time constants for procainamideinduced changes in VmaxIn vitro, and significantly longer than values obtained with a monoexponential model. Drug-induced changes in QRS duration were highly correlated with simultaneous changes measured by epicardial mapping (r=0.95, p<0.001), indicating that QRS duration is a valid index of drug effects on ventricular conduction. We concluded that procainamide causes interval-dependent changes in ventricular conduction in vivo that are consistent with a proportional relation between phase 0 inward current and the square of conduction velocity. These observations have important potential implications for the dosedependent and heart rate-dependent effects of antiarrhythmic drugs.