Study objective - The aim was to investigate whether heterogeneous coronary blood flow is mal-distributed during endotoxin shock.Design - Variables were studied before (t = 0) and at t = 90 and t = 120 min after bolus injection of saline (n = 6) or endotoxin (n = 6).Subjects - 12 anaesthetised mongrel dogs, weight 20-27 kg, were used.Measurements and main results - We studied myocardial blood flows in small tissue sections (of about 1 g in left and 2 g in right ventricle) with radioactive microspheres, together with haemody-namic variables and global myocardial metabolism. At t = 0 min in controls, regional flows per 100 g were heterogeneous and ranged from a factor 0.2 to 2.7 and 0.6 to 1.6 of mean flow per 100 g to the left and right ventricle respectively; heterogeneity was unchanged at t = 90 and t = 120 min. Between t = 0, t = 90, and t = 120 min regional flows correlated: r = 0.78(SD 0.14), n = 18, for left ventricle, and r = 0.70(0.17) for right ventricle. In the endotoxin group, cardiac output and mean arterial pressure decreased by 44(7) and 48(11)% respectively, and lactate increased by 3.2(0.6) mmol·litre -1 at t = 120 min. Global left ventricular blood flow and delivery and metabolism of 0 2 were unchanged; lactate extraction and external work fell. The ratio between global right ventricular 0 2 delivery and external work also rose. Regional blood flows ranged from a factor 0.2 to 2.7 and 0.1 to 1.8 of mean flow to left and right ventricles respectively; heterogeneity did not differ from controls and did not change with time. Flow correlations with time were reduced: 0.45(0.24) for left ventricle and 0.45(0.26) for right ventricle (both n = 18, p<0.005 v controls). The left ventricular endocardial to epicardial flow ratio fell; flow was redistributed to both layers.Conclusions - Heterogeneous blood flow is redistributed throughout the heart during canine endotoxin shock so that, at unchanged global blood flow and flow heterogeneity, flow decreases in some but increases in other areas. Flow maldistribution may be associated with focal ischaemia, which may be masked by a rise in O 2 uptake for a given workload (contractile inefficiency) in overperfused areas, and may thereby contribute to a fall in global myocardial external work for a given O 2 delivery.