We investigated the pathophysiology of hypoglycaemia in severe malaria in African children, especially the potential importance of glycerol as a substrate for gluconeogenesis, and whether substrate limitation contributes to hypoglycaemia in severe disease. Of 171 children with moderate or severe malaria, 16% were hypoglycaemic on admission, while at least 9% of children with severe malaria treated with quinine and a concurrent 4% dextrose infusion had a definite episode of hypoglycaemia after admission. Blood levels of gluconeogenic precursors are as high (alanine and lactate) or higher (glycerol) in those with either hypoglycaemia on or after admission as they are in children never having an episode of hypoglycaemia. Among children with severe malaria, however, those having a definite episode of hypoglycaemia at some stage are more acidotic and have greater evidence of renal impairment than those who are never hypoglycaemic (mean base excess -14.4 vs. -7.2, p < 0.001, mean creatinine 97 vs. 64, p < 0.001 and mean urea 8.1 vs. 5.8, p = 0.03, respectively). These data do not support a role for reduced gluconeogenic substrate supply in the pathogenesis of hypoglycaemia in severe childhood malaria, but do support the hypothesis that gluconeogenesis is impaired. Commonly-used bedside blood glucose monitoring devices may overestimate blood glucose measurements in the normal range, and paradoxically may also seriously overestimate the frequency of hypoglycaemia.