BACKGROUND A variety of previous studies have demonstrated reduced diastolic potential and electrical activity in atrial specimens from patients with heart disease. Although K+ channels play a major role in determining resting membrane potential and repolarization of the action potential, little is known about the effects of preexisting heart disease on human atrial K+ channel activity. RESULTS We characterized the inwardly rectifying K+ channel (IKI) and the muscarinic K+ channel [IK(ACh)] in atrial myocytes isolated from patients with heart failure (HF) and compared electrophysiological characteristics with those from donors (control) by the patch-clamp technique. Resting membrane potentials of isolated atrial myocytes from HF were more depolarized (-51.1 +/- 9.7 mV, mean +/- SD, n = 30 patients) than those from donors (-73.0 +/- 7.2 mV, n = 4 patients, P < .001). The action potential duration in HF was longer than that in donors. Although acetylcholine (ACh) shortened the action potential, reduced the overshoot, and hyperpolarized the atrial cell membrane in HF, these effects were attenuated compared with those observed in donors. The whole-cell membrane current slope conductance in HF was small, the reversal potential was more positive, and the sensitivity to ACh was less compared with donors. In single-channel recordings from cell-attached patches, IK1 channel conductance and gating characteristics were the same in HF and donor atria. When ACh was included in the pipette solution, IK(ACh) was activated in both groups. Single-channel slope conductance of IK(ACh) averaged 42 +/- 3 pS (n = 28) in HF and 44 +/- 2 pS (n = 4) in donors, and mean open lifetime was 1.3 +/- 0.3 milliseconds (n = 24) in HF and 1.5 +/- 0.4 milliseconds (n = 4) in donors. These values were virtually identical in the two groups (not significantly different, NS), although both single IK1 and IK(ACh) channel densities were less in HF. Channel open probability of IK(ACh) was also less in HF (4.0 +/- 1.2%, n = 24) than in donors (6.8 +/- 1.1%, n = 3, P < .01). The concentration of ACh at half-maximal activation was 0.11 mumol/L in HF and 0.03 mumol/L in donors. In excised inside-out patches, IK(ACh) from HF required higher concentrations of GTP and GTP gamma S to activate the channel compared with donors. These results suggest a reduced IK(ACh) channel sensitivity to M2 cholinergic receptor-linked G protein (Gi) in HF compared with donors. CONCLUSIONS Atrial myocytes isolated from failing human hearts exhibited a lower resting membrane potential and reduced sensitivity to ACh compared with donor atria. Whole-cell and single-channel measurements suggest that these alterations are caused by reduced IK1 and IK(ACh) channel density and reduced IK(ACh) channel sensitivity to Gi-mediated channel activation in HF.