To examine factors determining the haemodynamic and metabolic responses to treatment of diabetic ketoacidosis with alkali, groups of anaesthetised and ventilated rats with either diabetic ketoacidosis (mean arterial pH 6.86-6.96, mean arterial blood pressure 63-67 mm Hg) or hypovolaemic shock due to blood withdrawal (mean pHa 7.25-7.27, mean arterial blood pressure 36-41 mm Hg) were treated with sodium chloride ('saline'), sodium bicarbonate or 'Carbicarb' (equimolar bicarbonate plus carbonate). In the diabetic ketoacidosis series, treatment with either alkali resulted in deterioration of mean arterial blood pressure and substantial elevation of blood lactate, despite a significant rise in myocardial intracellular pH determined by 31P-magnetic resonance spectroscopy. These effects were accompanied by falling trends in the ratios of myocardial phosphocreatine and ATP to inorganic phosphate. Erythrocyte 2,3-bisphosphoglycerate was virtually absent in animals with diabetic ketoacidosis of this severity and duration. In contrast, in shock due to blood withdrawal, infusion of saline or either alkali was accompanied by a transient elevation of mean arterial blood pressure and no significant change in the already elevated blood lactate; erythrocyte 2,3-bisphosphoglycerate was normal in these animals. The effect of alkalinization in rats with severe diabetic ketoacidosis was consistent with myocardial hypoxia, due to the combination of very low initial erythrocyte 2,3-bisphosphoglycerate, alkali-exacerbated left shift of the haemoglobin-oxygen dissociation curve and artificial ventilation. No evidence was found for any beneficial effect of 'Carbicarb' in either series of animals; 'Carbicarb' and sodium bicarbonate could be deleterious in metabolic acidosis of more than short duration.