Single channel cell-attached patch and whole-cell clamp experiments on the mode of action of the K+ channel opener (KCO), levcromakalim, were performed in guinea pig isolated portal vein cells. At +20 mV (135/23 mM K+ in bath/pipette), 10 microM levcromakalim activated K+ channels with a chord conductance of 23.2 pS (K(KCO)), which were sensitive to the blocker of ATP-dependent K+ channels (K(ATP)), glibenclamide. Voltage steps from -80 mV to +20 mV activated 4-aminopyridine-sensitive K+ channels of 6.5 pS with properties of delayed rectifier K+ channels (Kv). In patches which upon a previous voltage step had revealed the existence of Kv, levcromakalim reduced the open-probability of Kv, but it did not concomitantly activate K(KCO). During the course of the experiments, but unrelated to the presence of levcromakalim, large conductance K+ channels (BK(Ca)) appeared which could be inhibited by iberiotoxin, a selective blocker of BK(Ca), and by the membrane-permeant calcium buffer, BAPTA/AM, but not by glibenclamide. Whole-cell current-voltage (i-V) relations were established in response to voltage ramps from +50 mV to -100 mV; on subtraction of control i-V curves from i-V curves obtained in the presence of 10 microM levcromakalim, the KCO-induced K+ current remained which was proportional to voltage. This is not compatible with the upward-bent curvature predicted by the GHK current equation for purely resistive channels at high [K+]i versus low [K+]o. In conclusion, in the guinea pig portal vein cells, no evidence could be established for the hypotheses that KCOs may act via conversion of Kv to K(ATP) (Beech and Bolton 1989; Edwards et al. 1993) or by activation of BK(Ca) (Balwierczak et al. 1995). In these cells, mild inward rectification of the levcromakalim-induced current was observed which underlines their relationship to K(ATP) in other tissues.