The deactivation kinetics of the delayed rectifier potassium current (IK) were studied in guinea-pig isolated ventricular myocytes at 35-36 degrees C using 5 microM E4031 to selectively block the rapidly activating component (IKr) and 300 microM thiopentone to selectively suppress the slowly activating component (IKs). IK was activated by depolarization of 400 or 1000 ms duration to +40 mV, and tail currents were analysed on repolarization to potentials between -30 and -60 mV. Before exposure to drugs, the deactivation of total IK was fitted by two exponential functions at all potentials and, at -40 mV, the time constants were 188 +/- 12 and 2510 +/- 185 ms (n = 25); increasing the pulse duration from 400 to 1000 ms (expected to increase IKs rather than IKr under these conditions) caused a significant increase in the amplitude of only the fast component of deactivation at -40 mV (from 372 +/- 32 to 479 +/- 46 pA). The decay of IKr was biexponential at all potentials when detected as the E4031-sensitive current and at -30 to -50 mV when detected as the thiopentone-resistant current and was accelerated as the membrane potential was made more negative. The amplitudes of the two components of decay of IKr (as either E4031-sensitive or thiopentone-resistant current) were similar and neither were significantly increased when pulse duration was increased from 400 to 1000 ms. The decay of IKs detected as the E4031-resistant current was biexponential at -30 mV and, in some cells, at -40 mV, but decayed with predominately a single, fast component of deactivation at -50 and -60 mV, which was accelerated at the more negative potentials. A slow component of decay of IKs was detected in more cells when thiopentone was used to isolate IKs, although it was of much smaller amplitude than the fast component of decay. The presence of a slow component of decay of the thiopentone-sensitive IKs may result from a slight block of IKr by thiopentone. The amplitude of only the fast decay phase of IKs (detected using either drug) was increased when pulse duration was lengthened from 400 to 1000 ms. It seems likely that the relative importance of IKr and IKs in rate-dependent shortening of action potential duration may need to be re-evaluated in view of the relatively rapid deactivation of IKs, together with the prominence of the slow component of deactivation of IKr compared with that of IKs.