Aim: To study the effect of doxapram on the frequency of apnoea, bradycardia and hypoxaemia. Methods: Fifteen infants, median gestational age at birth 27 weeks (range 24–30), age at study 27 days (12–60), with ≥6 episodes of bradycardia or hypoxaemia/6 h despite serum caffeine levels in the therapeutic range, received doxapram either intravenously (0.5–2 mg/kg/h) or orally (2–8 mg/kg every 2 h). Six-hour recordings of pulse oximeter saturation (SPO2), pulse waveforms, ECG, breathing movements and nasal airflow were performed immediately before as well as 1, 3 and 6 days after onset of treatment. Recordings were analysed for apnoea (≥4 s), bradycardia (heart rate < 2/3 of baseline) and hypoxaemia (SPO2 ≤80%). Results: There was no difference between enteral and intravenous administration; results are therefore presented for the total group. Doxapram resulted in a significant decrease in the frequency of apnoea [22 (11–27) vs. 14 (7–23)/h, p < 0.01], bradycardia [3 (0–7) vs. 1 (0–3)/h, p < 0.01] and hypoxaemia [8 (0–18) vs. 2 (0– 17)/h, p < 0.01] already after 1 day of treatment, which was sustained throughout the 6-day study period. Side effects included an increase in the proportion of time spent awake [5 (0–24) vs. 12% (3–28), p < 0.01] and in gastric residuals [0% of feeding volume (0–5) vs. 4% (0–19), p < 0.05]. Enteral was switched to intravenous doxapram in 3 of 9 infants because of gastrointestinal side effects. Conclusion: Doxapram substantially reduced the frequency of apnoea, bradycardia and hypoxaemia in these patients with caffeine-resistant apnoea of prematurity. Enteral administration, however, was not tolerated in a significant proportion (33%) of infants.