We investigated whether red cell 2,3-diphosphoglycerate (2,3-DPG) concentrations are reduced in critical illness, whether acidaemia, hypophosphataemia or anaemia influence 2,3-DPG, and whether there is any net effect on in vivo P50. Twenty healthy, non-smoking, male volunteers were compared with 20 male intensive care patients with APACHE 2 scores >20 on the preceding day. Those transfused in this time were excluded. Venous red cell 2,3-DPG concentrations were measured in both groups. In the patient group, routine multichannel biochemical profile and arterial blood gas analysis were also performed and in vivo P50 calculated. The mean 2,3-DPG concentration was significantly lower in the patient group than in the controls (4.2±1.3 mmol/l vs 4.9±0.5 mmol/l, P=0.016). The patients were well oxygenated (lowest arterial PO 2 =75 mm Hg) and showed a tendency to acidaemia (median pH 7.37, range 7.06 to 7.48) and anaemia (median haemoglobin concentration 113 g/l, range 89 to 154 g/l). By linear regression of patient data, pH had a significant effect on 2,3-DPG concentrations (r=0.6, P=0.011). Haemoglobin and phosphate concentrations did not, but there were few abnormal phosphate values. There was no correlation between 2,3-DPG concentrations and in vivo P50 (r 2 ≤ 0.08). We conclude that 2,3-DPG concentrations were reduced in a broad group of critically ill patients. Although this would normally reduce the P50, the reduction was primarily linked with acidaemia, which increases the P50. Overall, there was no net effect on the P50 and thus no affinity-related decrease in tissue oxygenation.