The present study examines the errors of measurement under working conditions using a capnograph and suggests a methodology to obtain optimal results in a given clinical situation. The authors compare the PCO2 measured with the aid of capnograph using CO2 absorption by infrared, with simultaneous measurements of PCO2 using a mass spectrometer as the reference. The observations measured included static and dynamic responses to a step variation of 5% CO2 and also the PACO2 in normal 9 to 12 year old children. The results show that the errors using a capnograph may reach 45 per cent at the highest respiratory frequencies. The influence of the dimensions of the sampling apparatus and the output of the sampling pump on the measurement of PACO2 are discussed. The static and dynamic calibration allow optimal operating conditions for the requirements of a patient in bed. (In this study children aged 9 to 12, with a respiratory frequency of less than 40/min). The errors after achieving optimal conditions is independent of respiratory frequency and always remained less than 5 per cent. A few simple rules are suggested to avoid the errors we have seen and if proper precautions are taken capnography can be considered as a good method for measuring PACO2 in children.