It is necessary, especially in neonatal work, for investigators to measure accurately the frequency response and common mode gain of respiratory pressure and flow transducers in air and other gas mixtures. Many of the systems designed for this task have been incompletely analysed and have unknown frequency responses themselves. The authors analyse aspects of systems employed previously and show that substantial amplitude and phase errors may have occurred. They describe a plethysmograph-based system which operates in any available gas mixture. A mathematical model of the acoustic microphone used as reference transducer, the microphone preamplifier, and the thermal behaviour of the plethysmograph, is developed to quantify the frequency response of the system. Maximum deviation from a perfect response is less than 1% in amplitude and 1 degrees in phase for both pressure and flow measurements over the range 1-100 Hz. Measurements using a pressure transducer mounted on the tip of a catheter indicate that the error due to amplitude and phase variation in the plethysmograph is less than 0.1 dB and 0.4 degrees at 100 Hz.