Bedside estimation of pulmonary gas exchange efficiency may be possible from step changes in FiO2 and subsequent measurement of arterial oxygenation at steady state conditions. However, a steady state may not be achieved quickly after a change in FiO2, especially in patients with lung disease such as COPD, rendering this approach cumbersome. This paper investigates whether breath by breath measurement of respiratory gas and arterial oxygen levels as FiO2 is changed can be used as a much more rapid alternative to collecting data from steady state conditions for measuring pulmonary gas exchange efficiency. Fourteen patients with COPD were studied using 4–5 step changes in FiO2 in the range of 0.15–0.35. Values of expired respiratory gas and arterial oxygenation were used to calculate and compare the parameters of a mathematical model of pulmonary gas exchange in two cases: from data at steady state; and from breath by breath data prior to achievement of a steady state. For each patient, the breath by breath data were corrected for the delay in arterial oxygen saturation changes following each change in FiO2. Calculated model parameters were shown to be similar for the two data sets, with Bland–Altman bias and limits of agreement of −0.4 and −3.0 to 2.2 % for calculation of pulmonary shunt and 0.17 and −0.47 to 0.81 kPa for alveolar to end-capillary PO2, a measure of oxygen abnormality due to shunting plus regions of low ratio. This study shows that steady state oxygen levels may not be necessary when estimating pulmonary gas exchange using changes in FiO2. As such this technique may be applicable in patients with lung disease such as COPD.