BACKGROUND: We observed two preterm infants who developed severe hypokalaemia following doxapram. We therefore wished to review the possible association between doxapram and severe hypokalaemia. STUDY DESIGN: A retrospective study of preterm infants born before 32 weeks of gestation and hospitalised in our intensive care unit in 2004. For each infant, treatment with doxapram or with any drug known to interfere with potassium metabolism, potassium intakes and episodes of hypokalaemia have been recorded. RESULTS: Out of 105 infants, 54 received doxapram. Doxapram-treated infants were significantly younger and had a lower birth weight. Doxapram treated infants were more likely to receive caffeine, furosemide, insulin and mechanical ventilation. There was no difference between the two groups for the other parameters. Hypokalaemia was frequently encountered in our population since it occurred in 76% of enrolled patients and severe hypokalaemia (potassium plasma level below 3 mmol/l) was found in 41%. Bivariate analysis underlined several risk factors for severe hypokalaemia: use of doxapram, gestational age below 28 weeks, use of mechanical ventilation, furosemide, ibuprofen, insulin and postnatal corticosteroids. Cox model's multivariate analysis showed that administration of furosemide and doxapram significantly increased the occurrence of severe hypokalaemia with relative risks of 4.9 (95% CI 1.9 to 12.5) and 8.2 (95% CI 3.1 to 21.7), respectively. CONCLUSIONS: This retrospective study underlines the high incidence of severe hypokalaemia in very preterm infants and an increased risk of severe hypokalaemia during doxapram treatment. We recommend potassium monitoring during any use of doxapram.