The speed of most metabolic processes is reduced to half by each 10 o C reduction in temperature. Therefore we examined the relation between safe ischaemic duration and temperature in a working, rat heart model perfused with erythrocyte suspension. We used a 30 mM K + concentration to arrest the heart and exclude the effect of contractile workload. The length of the ischaemic insult at each temperature was extended until post-ischaemic recovery of left ventricular output was reduced to 70-90% of control. The results show that when the arrested heart was cooled to 18 o C and 28 o C, the tolerable ischaemic duration was 60 and 40 minutes, respectively. At normothermia, the critical length of the ischaemic insult was 15-20 min. This means that the time to cause the same amount of damage is not doubled by each 10 o C reduction in temperature. This may be due to complex interactions between damage and protective mechanisms. The effect of subjecting hearts to ischaemia at normothermia appears to be much more variable than at hypothermia, as the endpoint functional recovery is very sensitive to slight changes occurring during the experiment.