It is well known that ischaemic preconditioning delays infarct size during regional ischaemic insults. However, the extent of this protective effect against different ischaemia periods has not been established, and any reduction in stunning has been difficult to demonstrate with regional models. In this study we have investigated ischaemic preconditioning in a buffer-perfused isolated rabbit heart model with a global ischaemic insult, and measured both infarct volume and functional recovery. Experiments were performed with three ischaemia time periods of 15, 20 and 30 min at 37 degrees C. Infarct volume (expressed as a percentage of left ventricular volume) was measured by tetrazolium staining after 2 hours reperfusion, and left ventricular developed pressure with an intraventricular balloon. Hearts preconditioned with 5 min ischaemia and 10 min reperfusion were compared with a control group. In this model, preconditioning resulted in a 57% reduction in infarct volume compared with control hearts (P = 0.02) subjected to 20 min of global ischaemia, but the degree of this infarct delaying effect was dependent on the ischaemia time and was only 37% (P = 0.02) and 11% (N.S.) with a 30 min and 15 min ischaemic challenge respectively. Recovery of post-ischaemic left ventricular developed pressure as a percentage of the pre-ischaemic value correlated very well with infarct volume in control r = -0.82 (P < 0.001) and preconditioned r = -0.78 (P < 0.001) groups, and the slope of the regression lines was similar for both groups. These results demonstrate that the degree of protection produced by preconditioning is not uniform but varies with the length of the ishaemic insult. By measuring both infarct volume and functional recovery we have been able to confirm that any post-ischaemic improvement in global left ventricular function produced by preconditioning is secondary to reduced infarction, and hence that preconditioning does not attenuate stunning.