Regional cerebral blood flow and oxygen utilisation can be measured using positron emission tomography and the continuous inhalation of 15O-labelled carbon dioxide and molecular oxygen. However, there is concern about the propagation of errors in this technique. This arises from statistical uncertainties, inherent in the transmission and emission scans, being amplified in the steady state model used for calculating the physiological parameters. The magnitude of this effect has been studied using a series of repeat transmission and emission phantom measurements in which pixel count densities equivalent to those seen in the clinical data were recorded. These measurements have been used to determine the final propagated errors that occur in the calculated values of regional cerebral blood flow, oxygen extraction fraction, and oxygen utilisation rate.