ObjectiveTo measure within-subject changes in ventilation/perfusion (V′/Q′) mismatch in response to a protocol of individualised nasal continuous positive airway pressure (CPAP) level selection.DesignSingle-arm, non-randomised, feasibility trial.SettingThree centres in the Children’s Hospital of Philadelphia neonatal care network.PatientsTwelve preterm infants of postmenstrual age 27–35 weeks, postnatal age >24 hours, and receiving a fraction of inspired oxygen (FiO2) >0.25 on CPAP of 4–7 cm H2O.InterventionsWe applied a protocol of stepwise CPAP level changes, with the overall direction and magnitude guided by individual responses in V′/Q′ mismatch, as determined by the degree of right shift (kilopascals, kPa) in a non-invasive gas exchange model. Best CPAP level was defined as the final pressure level at which V′/Q′ improved by more than 5%.Main outcome measuresWithin-subject change in V′/Q′ mismatch between baseline and best CPAP levels.ResultsThere was a median (IQR) within-subject reduction in V′/Q′ mismatch of 1.2 (0–3.2) kPa between baseline and best CPAP levels, p=0.02. Best CPAP was observed at a median (range) absolute level of 7 (5–8) cm H2O.ConclusionsNon-invasive measures of V′/Q′ mismatch may be a useful approach for identifying individualised CPAP levels in preterm infants. The results of our feasibility study should be interpreted cautiously and replication in larger studies evaluating the impact of this approach on clinical outcomes is needed.Trial registration numberNCT02983825.