Objective Metabolic acidosis in the preterm infant is a frequent occurrence and often treated with fluid and buffer boluses. Our aim was to identify the origin of metabolic acidosis in preterm infants, using Stewart’s method of strong ion difference.

Method Data were collected from blood gases, electrolytes, and albumin from preterm infants during their first week of life. These clinically indicated samples were examined for causes of acidosis using anion gap corrected for albumin (AGcorr), and Stewart’s strong ion method.

Results 102 blood gas samples were collected from 12 preterm infants (gestation 26–31 weeks). Of these, 90(88.2%) had a pH<7.35. PCO₂ was elevated in 94/102 (92%) samples. Hyperchloraemia (>107 mmol/l) was present in 76/102(74.5%) of blood samples. Raised AGcorr occurred in 75/96(78%) available samples. Raised lactate (>2 mmol/l) occurred in 14/102 samples (14%).

In 26 samples, simultaneous measurement of electrolytes allowed us to calculate total tissue acidosis and unmeasured anions (UMA). Tissue acidosis occurred in 23/26 (88%) samples. This was due solely to increased UMA (>3 mEq/l) in 16/23 (69%) samples and solely to increased lactate in only one case. Mixed tissue acidosis (raised UMA and lactate) was present in 6/23 (26%) of samples. In addition 19/26 (73%) samples were hyperchloraemic.

Conclusions Metabolic acidosis in preterm infants is complex and is most frequently caused by a combination of unmeasured anions and hyperchloraemia. Isolated lactic acidosis is infrequent. The use of pH and base deficit as surrogates for lactate is imprecise.