The ventricular action potential (AP) shortens exponentially upon a progressive reduction of the preceding diastolic interval. Steep electrical restitution slopes have been shown to promote wavebreaks, thus contributing to electrical instability. The present study was designed to assess the predictive value of electrical restitution in hypokalemia-induced arrhythmogenicity. We recorded monophasic APs and measured effective refractory periods (ERP) at distinct ventricular epicardial and endocardial sites and monitored volume-conducted ECG at baseline and after hypokalemic perfusion (2.5 mM K(+) for 30 min) in isolated guinea pig heart preparations. The restitution of AP duration measured at 90% repolarization (APD(90)) was assessed after premature extrastimulus application at variable coupling stimulation intervals, and ERP restitution was assessed by measuring refractoriness over a wide range of pacing rates. Hypokalemia increased the amplitude of stimulation-evoked repolarization alternans and the inducibility of tachyarrhythmias and reduced ventricular fibrillation threshold. Nevertheless, these changes were associated with flattened rather than steepened APD(90) restitution slopes and slowed restitution kinetics. In contrast, ERP restitution slopes were significantly increased in hypokalemic hearts. Although epicardial APD(90) measured during steady-state pacing (S(1)-S(1) = 250 ms) was prolonged in hypokalemic hearts, the left ventricular ERP was shortened. Consistently, the epicardial ERP measured at the shortest diastolic interval achieved upon a progressive increase in pacing rate was reduced in the hypokalemic left ventricle. In conclusion, this study highlights the superiority of ERP restitution at predicting increased arrhythmogenicity in the hypokalemic myocardium. The lack of predictive value of APD(90) restitution is presumably related to different mode of changes in ventricular repolarization and refractoriness in a hypokalemic setting, whereby APD(90) prolongation may be associated with shortened ERP.