Background: Classifying the severity of bronchopulmonary dysplasia (BPD) by continuous numerical variables would facilitate follow-up of disease progression and quantified analysis of disease determinants. Objectives: To non-invasively measure oxygenation impairment in BPD by the degree of right-to-left shunt, right shift of the oxyhaemoglobin dissociation curve and ventilation/perfusion (VA/Q) inequality and to explore their relation with clinical parameters. Methods: Prospective cohort study of 24 infants with a median (interquartile range, IQR) gestation of 25 weeks (24-27) and a birth weight of 0.70 kg (0.63-0.93), studied at 36 days (30-66), at a postmenstrual age (PMA) of 33 weeks (29-36). Inspired oxygen (FIO2) was varied to obtain three to five transcutaneous oxygen saturation (SpO2) values between 85 and 96%. Values of shunt, shift and VA/Q were obtained by plotting the paired data of SpO2 against FIO2 for each infant using a unique program. Right-to-left shunt, right shift of the oxyhaemoglobin dissociation curve and VA/Q were measured in infants born <32 weeks PMA receiving oxygen at 28 days. <b>Results: The median (IQR) shunt was 8% (0.3-16.5), shift 14.5 kPa (10.9-19.4) and VA/Q 0.40 (0.30-0.48). Shunt, shift and VA/Q were significantly related to gestational age (GA) at birth, PMA at study, weight at study and weight gain per week. Conclusions: Severity of pulmonary oxygenation impairment in BPD can be quantified at the cot-side by non-invasive measurement of shunt, shift and VA/Q. Low GA at birth, low weight at birth and at the time of study and impaired weight gain are significantly associated with the severity of oxygen-exchange impairment in infants with BPD.