The system mean transit time (Tsys)of the impulse response function describing contrast material transit through the coronary circulation was determined from serial digital angiographic images. The linearity, reproducibility, and relations with regional myocardial perfusion and conventional time-density curve parameters, time to peak concentration (TPC), and exponential washout rate (k) were assessed in a dynamic flow x-ray phantom (n=46) and in six open-chest dogs (/t=102) while coronary flow was altered by stenosis and/or hyperemk stimuli. In the phantom studies, the inverse of the system mean transit time (T±,-1) closely predicted flow/volume (r=0.99, slope=0.99). In dogs, Tsys,-1was independent of the shape of the contrast bolus injection (single or double-peaked), class of contrast agent (ionic or nonionic), the type of hyperemic stimulus (dipyridamole, dipyridamole plus norepinephrine, transient total occlusion, or ionic contrast media), and was highly reproducible between adjacent myocardial regions served by the same artery (r=0.98±0.01). There was a strong correlation between Tsys,-1and regional coronary flow for stenotic and/or hyperemic vessels (r=0.94, distribution volume=14.9 ml/100 g) over a wide range (0-514 ml/min/100 g). Tsys,-1performed better than conventional time-density curve parameters TPC-1and k for predicting phantom flow/volume ratios and regional myocardial blood flow in the dog. These data suggest that both digital coronary angiography and coronary contrast transit can be modeled as linear systems and that impulse response analysis may provide accurate and reproducible estimates of regional myocardial blood flow.