1. Single sodium channel currents were recorded from canine ventricular myocytes in cell‐attached patches. The relative rates of single‐channel activation vs. inactivation as well as the voltage dependence of the rate of open‐channel inactivation were studied. 2. Ensemble‐averaged sodium currents showed relatively normal activation and inactivation kinetics, although the mid‐point of the steady‐state inactivation (h infinity) curve was shifted by 20‐30 mV in the hyperpolarizing direction. This shift was due to the bath solution, which contained isotonic KCl to depolarize the cell to 0 mV. 3. Steady‐state activation showed less of a voltage shift. The threshold for eliciting channel opening was around ‐70 mV and the mid‐point of activation occurred near ‐50 mV. 4. The decline of the ensemble‐averaged sodium current during a maintained depolarization was fitted by a single exponential function characterizing the apparent time constant of inactivation (tau h). The apparent rate of inactivation was voltage dependent, with tau h decreasing e‐fold for a 15.4 mV depolarization. 5. The relative contributions of the rates of single‐channel activation and inactivation in determining the time course of current decay (tau h) were examined using the approach of Aldrich, Corey & Stevens (1983). Mean channel open time (tau o) showed significant voltage dependence, increasing from 0.5 ms at ‐70 mV to around 0.8 ms at ‐40 mV. At ‐70 mV tau h was much greater than tau o, while at ‐40 mV the two time constants were similar. 6. The degree to which the kinetics of single‐channel activation contribute to tau h was studied using the first latency distribution. The first latency function was fitted by two exponentials. The slow component was voltage dependent, decreasing from 19 ms at ‐70 mV to 0.5 ms at ‐40 mV. The fast component (0.1‐0.5 ms) was not well resolved. 7. Comparing the first latency distribution with the time course of the ensemble‐averaged sodium current at ‐40 mV showed that activation is nearly complete by the time of peak inward sodium current. However, at ‐70 mV, activation overlaps significantly with the apparent time course of inactivation of the ensemble‐averaged current. 8. Using the methods of Aldrich et al. (1983) we also measured the apparent rate of open‐channel closing (a) and open‐channel inactivation (b). Both rates were voltage dependent, with a showing an e‐fold decrease for an 11 mV depolarization and b showing an e‐fold increase for a 30 mV depolarization.(ABSTRACT TRUNCATED AT 400 WORDS)