One hundred and six critically ill adult patients were monitored continuously with a transcutaneous oxygen sensor (PtcO2); they also were intermittently monitored with conventional invasive hemodynamic and oxygen transport variables. A total of 1073 data sets were taken on 41 patients in the ICU and 65 patients in the operating room. The patients were divided into three groups by cardiac index (CI): relatively normal flow, CI greater than 2.2 L/min x M2; moderate low flow shock, 2.2 greater than CI greater than 1.5 L/min x M2; and severe low flow shock, CI less than 1.5 L/min x M2 x PtcO2 and arterial oxygen tension (PaO2) were compared in two ways: first by linear regression and second by a more simple clinical guide by indexing each transcutaneous value by its respective arterial value (PtcO2 index = PtcO2/PaO2). For 934 data sets taken on 92 patients not in shock, there was a correlation coefficient (r) of 0.89 and a PtcO2 index 0.79 +/- 0.12 (SD). In 5 patients with moderate shock, the r was 0.78 and the PtcO2 index was 0.48 +/- 0.07. In 9 patients with severe shock, there was no correlation between PtcO2 and PaO2 and the PtcO2 index was 0.12 +/- 0.12. In all cases of severe shock, the PtcO2 value responded quickly to changes in blood flow with an approximate 1 min response time (95%). The patients not in shock responded to changes in inspired oxygen concentration (FIO2) with changes in PaO2 and PtcO2 values; the 95% response time was approximately 2 min. The authors conclude that the normal value for PtcO2 for adult surgical patients who are hemodynamically stable is 79 +/- 12% of the PaO2 and that PtcO2 values were reliable, continuous, noninvasive trend monitors of PaO2 in these patients. During circulatory problems when PtcO2 values were compared to PaO2 values (PtcO2 index), the changes reflected trends in the severity of low flow shock.