In studies of preconditioning, a variety of models have been used. The aim of the present study was to find the optimal preconditioning model for preservation of cardiac function during reperfusion of globally ischaemic, Langendorff-perfused rat hearts. Cardiac function was assessed by the occurrence of severe reperfusion arrhythmias (ventricular fibrillation or asystolia), heart rate (HR), left ventricular systolic (LVSP), end diastolic (LVEDP), and developed pressures (LVDP = LVSP - LVEDP), as well as coronary flow (CF). Series 1 (n = 17) in each group: control perfusion for 20 min without preconditioning or 2 episodes of 2, 3, 4, or 5 min of ischaemia, each followed by 5 min reperfusion, before 25 min ischaemia and 60 min reperfusion. Preconditioning reduced the incidence of reperfusion arrhythmias, attenuated the reperfusion-induced increase of LVEDP, and increased CF, but did not influence LVSP, LVDP, or rate x pressure-product (RPP = LVSP x HR) during reperfusion. The greatest effect was found by 2 min ischaemia and 5 min reperfusion. In series 2 (n = 17 in each group) control perfusion for 7 or 28 min, or preconditioning with 1-4 episodes of 2 min ischaemia and 5 min reperfusion before 35 min ischaemia and 60 min reperfusion were compared. Reduction of severe reperfusion arrhythmias and LVEDP elevation, as well as improvement of CF, LVDP, and HR in preconditioned hearts were observed in series 2. Optimal cardioprotection was achieved by only one episode of preconditioning. In conclusion, preconditioning before global ischaemia improved cardiac function during reperfusion of isolated rat hearts. The most marked effects were reduction of severe reperfusion arrhythmias and attenuation of diastolic dysfunction. Although all preconditioning models employed were cardioprotective, 1 episode of 2 min ischaemia provided optimal protection.