In some angiographic methods for measurement of mean coronary flow in ml/min, a threshold is applied to 'concentration-distance' curves obtained from a constant rate injection by computing the intravascular contrast medium concentration along the main coronary branches. If the shape of the velocity profile would remain parabolic throughout the cardiac cycle, the correct threshold value would be '50% of the concentration at the injection site'. But, coronary flow being strongly pulsatile, the shape of the velocity profile must be expected to vary appreciably within the cardiac phase. In order to investigate if a single, appropriate threshold value nevertheless exists for a great variety of coronary flow pulses and velocity profiles, the spreading of contrast medium injected continuously in a tube perfused by a time varying flow Q(t) was studied by computer simulation. While the particular time courses of flow and velocity profile appear to be of secondary importance, the ratio 'injection rate to peak coronary flow' has a major impact. If it is equal to or greater than 1, a threshold value of 47% is the best choice. If the ratio is markedly less than 1, no appropriate threshold exists and use of the 47% threshold will result in considerable flow underestimations. This was fully confirmed by measurements of absolute coronary flow performed in patients.