Transcutaneous oxygen tension (tcPO2) measured with a heated electrode was compared with arterial oxygen tension (PaO2) in three groups of adult patients-group A: 20 patients with chronic respiratory disease; group B: eight hypothermic patents studied immediately after cardiopulmonary bypass surgery; and group C: 14 patients in an intensive care unit who had in all 35 comparisons of tcPO2 and PaO2. For group A, the relationship of the two methods is given by the regression equation (in mmHg): tcPO2 = 0.88 PaO2 + 5.0 (standard error = 5.2: r = 0.93: p less than 0.001) and for group C tcPO2 = 0.69 PaO2 + 14.9 (standard error =12.0, r = 0.90: p less than 0.001). A method of in vivo calibration using a single arterial blood sample improved the reliability of the method in group C. In the group B patients there was no correlation between tcPO2 and PaO2 even when supplementary heating was applied to the skin. Continuous monitoring can reveal large fluctuations in arterial PO2 which would be missed by the use of intermittent arterial samples. The transcutaneous electrode can be employed usefully in intensive care monitoring of adult patients and in physiological studies where repeated arterial samples would otherwise be required.