Sampling arterialized earlobe blood is thought to be easier and less painful than direct arterial puncture, and to allow measurement of blood gas values during exercise without the need to insert an arterial cannula. However, arterialized earlobe oxygen tension (PO2) often underestimates arterial PO2 at rest, and is not fully validated during exercise. We have therefore conducted a prospective study to compare values of PO2 and carbon dioxide tension (PCO2) and the discomfort experienced by adult subjects undergoing the two methods of blood sampling during exercise. Seventy consecutive adult patients were studied. Blood samples were drawn simultaneously from the radial artery and arterialized earlobe of each patient during the last minute of an 8 min exercise. Values of PO2 and PCO2 were measured by means of blood gas electrodes. The correlation coefficients between the two samples were 0.92 for PO2 and 0.91 for PCO2. However, the bias and the limits of agreement between the two methods were wide for PO2 (mean+/-2SD of the differences between the two methods: 0.63+/-1.50 kPa (4.7+/-11.2 mmHg)). The bias and the limits of agreement were smaller for PCO2. Patients felt that the earlobe method was not associated with less discomfort than radial artery puncture. We conclude that arterialized earlobe blood oxygen tension is not a good substitute for arterial oxygen tension during exercise, and should not be used to assess arterial oxygen tension in adults during exercise.