The linear regression analysis of infarct size (IS) v ischemic myocardial blood flow (MBF) does not account for the heterogeneity of MBF and infarcted tissue; moreover, it cannot assess a blood flow threshold for infarction (MBFT) accurately, as with ischemic preconditioning (IP) the close relationship between ischemic MBF and IS otherwise observed is lost. Finally, the impact of resting blood flow on myocardial infarction cannot be considered in such analysis. Therefore, in a retrospective data analysis of 32 enflurane-anaesthetized swine undergoing 90 min severe ischemia and 120 min reperfusion without (CON, n = 12) or with IP induced by either 3 (IP3, n = 8) or 10 min ischemia (IP10, n = 12) and 15 min reperfusion, a MBFT was assessed by logistic regression (LR) in individual tissue pieces. MBFT was arbitrarily defined as that ischemic MBF (microspheres) at which infarct probability was 0.2, derived from the ratio of infarcted (n = 141, TTC) to all tissue samples (n = 684). The duration of the preconditioning ischemia and MBF both at rest and during the sustained ischemia were significant predictors of infarct probability. Ischemic MBFT at an infarct probability of 0.2, was 0.089 +/- 0.023 ml/min/g in CON. MBFT was decreased to 0.051 +/- 0.03 ml/min/g with IP3 (P < 0.05 v CON) and further to 0.004 +/- 0.037 ml/min/g with IP10 (P < 0.05 v CON, IP3). Corresponding to the leftward shift of MBFT, the relationships between infarct probability and MBF were shifted in parallel by IP with no change in their slopes.