Transcutaneous-PO2 (tc-PO2 (tc-PCO2) at 44 degrees C and transcutaneous-PCO2) at 38, 42, 43 and 44 degrees C were measured with a currently available monitoring system (TCM222, Radiometer, Copenhagen) in 64 newborn infants with severe respiratory insufficiency during the first five days of life. Tc-PCO2 at all four temperatures correlated better with arterial blood-PCO2 (aB-PCO2), than tc-PO2 with aB-PO2. However, the sensitivity and specificity of tc-PO2 and tc-PCO2 were similar with regard to maintaining aB-PO2 and aB-PCO2 within specified limits. Tc-PCO2 increased relatively with increasing electrode temperature by a factor which was similar to the anaerobic temperature coefficient of PCO2 in blood. The coefficient of variation of duplicate measurements was 10% for tc-PO2 and 5% for tc-PCO2. Electrode drift after an average of 3 hours patient monitoring was 2% +/- 6% (1 SD) for tc-PO2 and -3% +/- 6% for tc-PCO2. We conclude that tc-PO2 and tc-PCO2 are a valuable supplement to arterial blood gas measurements, but the variable correlation with arterial blood gas values and the electrode drifts which may occur, mean that they cannot fully replace arterial sampling.