BACKGROUND "Ideal" alveolar oxygen tension (PAO2) is a calculated entity and the alveolar-arterial oxygen tension difference (PA-aO2) is used to evaluate gas exchange function of the lungs. Accurate calculations of PAO2 necessitate measurements of the respiratory exchange ratio (RER), which is less frequently done, and most often approximations are made. The measured end-tidal oxygen tension (PETO2) is a reflection of the alveolar oxygen tension. The aim was to study the relationship between PAO2 and PETO2, and to see whether the end-tidal to arterial oxygen tension difference (PET-aO2) could give the same information about lung function as PA-aO2. METHODS Twenty patients admitted for cardio-pulmonary exercise tests were studied. They bicycled for 4 min at each work load until maximum work load was reached. Arterial blood gases were analysed before, after 4 min at each work load, at maximum work load and after 2 min of recovery. A metabolic computer measured mixed expired gas concentrations. End-tidal gas concentrations were measured with a side stream gas analyser. RESULTS We measured major increases in oxygen uptake, carbon dioxide elimination and RER. PAO2 and PETO2 increased at maximum exercise and during recovery. PAO2 and PETO2 were closely correlated during the study, through great changes in oxygen uptake and RER (r=0.88). When correction was made for wet gas the median difference was 0.12 kPa. CONCLUSIONS At ambient air (FIO2=0.21), PET-aO2 as a respiratory index may give equivalent information to PA-aO2, without the need for measurements of mixed expired gas tensions or the hazard of an assumed RER.