SummaryTo determine which warming system more closely approximates a neutral thermal environment, the oxygen consumptions of 16 premature babies <1500 g were measured in a convectively heated incubator and a radiantly heated incubator. Both systems were controlled to maintain a skin temperature of 36°C. The oxygen consumptions of the infants were not significantly different in the two incubators; thus, there was no detectable advantage to the use of convective or radiant energy in approximating a neutral thermal environment in an incubator for the small premature infant. That proportion of total heat loss from the babies due to radiant losses in the convective incubator was directly calculated from incubator temperature using equations described in the paper, and found to be 68 ± 3% SE.