Thin preparations of cat ventricular muscle were mounted in a single sucrose gap and superfused with Tyrode solution containing 1-5 microM-D600. In voltage-clamp experiments lasting for 40-180 min, stimulation with standard pulses (-50 to 0 mV, 300 ms) at 0.33 Hz depressed Ca-dependent slow inward current (ICa) to less than 20% of its pre-drug amplitude. A reproducible unblocking of ca. 75% of the blocked Ca channels could be achieved with a single hyperpolarizing pulse (90 s at -90 mV); stimulation (conditioning) at 0.33 Hz re-established full block within thirty pulses. The time and voltage dependence of block and unblock were examined by varying the frequency and duration of voltage-clamp pulses. The time course of unblock was usually monoexponential. The time constant was voltage dependent and declined from 9 min at -50 mV to 5 s at -110 mV. Block appears to depend on channel state, resting channels being highly resistant to block and open channels very susceptible. D600 also binds to inactivated channels but at a much slower rate than to open channels. A small U-shaped component of block was induced by conditioning to potentials between +10 and +80 mV. This block seemed to be unrelated to channel state, suggesting that drug binding may also be dependent on voltage. Quicker rates of block after repetitive conditioning, and slow wash-out of the drug, may indicate the existence of an intramembrane drug pool distinct from the primary pool in the intracellular fluid. The interaction of D600 with Ca channels is discussed in terms of a channel state model. In many respects this interaction resembles that of local anaesthetics with Na channels.