Much clinical and experimental evidence suggests that pancreatic ischaemia in the early phase of acute pancreatitis is important in the development of pancreatic necrosis. While depletion of intravascular volume has often been assumed to be the main circulatory defect, an additional disturbance of pancreatic microcirculation has been demonstrated experimentally. Possible contributory mechanisms include chemical-induced vasoconstriction, direct injury of vessel wall, intravascular coagulation and increased endothelial permeability resulting in pancreatic oedema, haemoconcentration and impaired venous drainage. Pancreatic ischaemia as a consequence of these local effects seems to be responsible for the transition of mild pancreatitis to parenchymal necrosis. In experimental models the beneficial effect of various drugs and of sympathetic blockade has been ascribed to an improvement in pancreatic perfusion. Although effective volume therapy is generally accepted as the mainstay of conservative treatment in acute pancreatitis, the efficacy of different fluid preparations is still controversial, and simple fluid resuscitation has not been shown to prevent the development of parenchymal necrosis. The specific impairment of pancreatic microcirculation cannot be prevented merely by replenishment of intravascular volume with crystalloids, albumin or plasma despite normalization of macrohaemodynamics. In contrast, partial replacement of blood by dextran preparations has been shown to increase pancreatic perfusion by improving blood fluidity. Isovolaemic haemodilution in conjunction with conventional fluid therapy may provide a new and effective means of protecting the pancreas from secondary injury due to the early ischaemic phase of acute pancreatitis.