Uncuffed endotracheal tubes (ET) are commonly used for mechanically ventilated infants to protect airways, but this bears the risk of an air leak around the ET. In contrast to the measurements of tidal volume and respiratory mechanics, very little is known about the effect of ET leaks on capnographic measurements. To investigate the relationship between ET leakage and the CO2 measuring error of the exhaled breathing gas a neonatal lung model was used consisting of two silicon bellows. A constant but very low flow of pure CO2 was injected so that the exhaled breathing gas contained an adjustable CO2 plateau. This lung model was ventilated via a 3 mm inner diameter ET with different respiratory rates (RR) (20, 40 and 60 min(-1)). ET leaks (0-80%) were simulated by a needle valve. The end-expiratory CO2 partial pressure (PetCO2) was reduced up to zero when an air leak was simulated. Provided that the exhaled CO2 reached a plateau before entering the CO2 washout of the sample chamber by the leak flow, the magnitude of the CO2 decreased independently of RR, and only slightly with increasing ET leak. For ET leaks of up to 20% the CO2 plateau error was <10%. However, in newborns with stiff lungs and a short alveolar plateau capnographic measurements should be interpreted with caution if the ET leak is considerable.