A non-invasive technique for open circuit calorimetry based on a Vickers 79 incubator as the gas collection chamber was developed and evaluated. The technique, which involved drawing air from the incubator hood at 11 1/min for gas analysis, did not have a significant cooling effect on the infant and the noise levels within the incubator did not exceed current safety standards. A new technique for checking the calibration of the whole system was developed as the traditional alcohol burn method proved unsatisfactory. Either pure oxygen or carbon dioxide were fed into the incubator hood at controlled physiological rates using a calibrated throttling valve. Over a number of calibration checks the mean error of the system proved to be +/- 4.3% for oxygen and +/- 4.45% for carbon dioxide. Less than a fifth of this error was attributable to the pump and flow meter. When oxygen is introduced to the system at a constant physiological rate it takes 30 min for a plateau to be reached. Therefore it is recommended that this system is used when readings are to be taken in a steady state situation or data is to be cumulated over a long period of time such as in an energy balance study. A run in period of 30 min before collecting data is essential. It is recommended that a calibration check is performed before each study.