Ten premature infants nursed on servocontrolled radiant warmer beds were studied in three environments designed to alter one or more factors affecting heat transfer (convection, evaporation, and radiation). In the control environment, infants were nursed supine on an open warmer bed. The second environment (walled chamber) was designed to reduce convection and evaporation by placing plastic walls circumferentially around the bed. In the third environment convection and evaporation were minimized by covering infants with a plastic blanket. Air turbulence, insensible water loss, and radiant warmer power were measured in each environment. There was a significant reduction in mean air velocity in the walled chamber and under the plastic blanket when compared to the control environment. A parallel decrease in insensible water loss occurred. In contrast, radiant power demand was the same for control and walled environments, but decreased significantly when infants were covered by the plastic blanket. This study suggests that convection is an important factor influencing evaporation in neonates nursed under radiant warmers. The thin plastic blanket was the most effective shield, significantly reducing radiant power demand.