More than 1% of infants are born premature. Many of these children require special treatment because of immature organs and body functions. CO2 is an important parameter to monitor in order to avoid serious brain damage. Blood sampling of CO2 has several shortcomings and non-invasive transcutaneous CO2 is being investigated in order to assess its potential to contribute with the same type of information as blood CO2 measurements. The present study assesses the quality of transcutaneous CO2 data by comparing it to the "golden standard" blood CO2 data, in order to provide clinicians with a better understanding of the usefulness and limitations of transcutaneous CO2 data in neonatal care. The study shows that for low transcutaneous CO2 the error is relatively high and in most cases the true CO2, represented by the blood CO2, which can be regarded as the "gold standard", is higher than the measured transcutaneous CO2. The opposite is the case for high transcutaneous CO2. It is discussed how this is not due to any systematic error in the equipment, but due to the natural behaviour of noisy data.