Blood sampling from the arterialized earlobe is widely used in clinical exercise testing but Fajac et al. (1998) (Eur. Respir. J. 11, 712-715) have shown that arterialized P(O2) Pc(CO2) is not a valid surrogate for Pa(O2). In the present study, in order to detect disturbances in pulmonary gas exchanges during clinical exercise testing from the alveolar-arterial gradient of P(O2) (P[Ai-a](O2)), a correction factor for Pc(O2) was validated from data on a large cohort (107 patients at one or two levels of exercise: 172 pairs of samples). Pulmonary gas exchanges and pH, P(O2), P(CO2), PA(iO2) and P(Ai-a)(O2) from arterial and arterialized blood were measured or computed. Arterial and arterialized pH and P(CO2) (and thus PA(iO2)) were similar but P(CO2) was lower than arterial P(O2) (Pa(O2)). However, when corrected for the systematic bias between Pa(O2) and Pc(O2), which increased with Pc(O2), Pc(O2) adequately detected disturbances in pulmonary gas exchanges with a very high sensibility and specificity (predictive values of a negative or positive test ∼95%).