Continuous transcutaneous oxygen (TcPo2) monitoring was used to study low-birth-weight infants (15 control, 15 experimental) during their stay in the intensive care nursery (ICN). Twenty hours of continuous recordings were made of heart rate, respiratory rate, and TcPo2 in the first five days after birth. Personnel caring for the control infants were blind to the TcPo2 measurements. Those caring for the experimental group were instructed in the operation of the TcPo2 monitor and used the device to modify the timing and extent of procedures to minimize "undersirable time" (TcPo2 less than 40 or greater than 100 torr). Control group infants had an average of 40 min/20 hr in "undesirable time." When personnel used TcPo2 monitoring to modify care this "undesirable time" was reduced to 6 min/20 hr. This difference was statistically significant (P less than .001). Infants in the control group were handled more frequently, and experienced more hypoxemia. Of the hypoxemic time suffered by the control infants 75% was associated with handling. Only 5% of the hypoxemia or hyperoxemia was detected by blood Pao2 analysis or conventional monitors. Intermittent arterial blood gas sampling is an inadequate method for monitoring arterial oxygenation in low-birth-weight infants since hypoxemia and hyperoxemia are often missed. Continuous TcPo2 monitoring improves care by diminishing "undesirable time."