Flow quantitation with nitrogen-13 ammonia ((13)NH(3)) and positron emission tomography (PET) is dependent on an accurate blood time-activity curve. This is conveniently derived from the PET images by drawing a region of interest in the left ventricular cavity. The blood time-activity curve obtained in this way, however, may contain spillover from the myocardial wall. The purpose of this study was to analyse the effect of wall to blood pool spillover. Additionally, we analysed the application of a left atrial input function. Using computer simulations, we investigated the effect of spillover from the myocardial wall to the left ventricular input function and the effect of time delay on the left ventricular input function. An oxygen-15 carbon monoxide PET study of seven normal volunteers was used to investigate possible recovery issues regarding the left atrial input function. Finally, (13)NH(3) studies of 31 normal volunteers during rest and dipyridamole stimulation were analysed using either a left atrial or a left ventricular input function. The simulation studies showed that myocardial wall to blood pool spillover causes a considerable underestimation of the regional blood flow values in hyperaemic flow studies. Neither time delay nor recovery issues prevent flow quantitation with a left atrial input function. The (13)NH(3) studies revealed no significant difference between the resting blood flow values, whereas the hyperaemic blood flow values were underestimated by 8% (P<0.01) on average (up to 40% individually) when using a left ventricular input function compared with a left atrial input function. Spillover of activity from the left ventricular wall to the blood time-activity curve is of importance in hyperaemic flow studies using (13)NH(3). Application of a left atrial input function is a possible solution to these issues.