Animal experimentation has shown that the instantaneous pressure-velocity of coronary blood flow in the hyperaemic phase has a linear relationship. The slope of this regression evaluates coronary reserve independently of haemodynamic variables and the X-intercept (zero flow pressure or Pf = 0) determines the intra-myocardial back pressure which could influence the regulation of coronary flow. The object of this study was to evaluate the instantaneous pressure-velocity relationship of coronary flow in clinical practice and to analyse the reproducibility of this parameter. Forty-nine patients were divided into two groups, depending on whether their coronary arteries were angiographically normal (n = 34) or atheromatous with stenosis > or = 35% of the reference diameter (n = 15). Recordings of coronary flow velocity were made with a Doppler transducer mounted on a 0.018 inch guide wire. The slope of the diastolic linear segment of the pressure-velocity relationship was determined at the peak of papaverine-induced vasodilation from 4 consecutive cycles by a regression analysis. The pressure value at 0 flow was obtained by extrapolation of the regression slope to the axis of aortic perfusion pressure. A good quality spectral recording allowing reliable analysis of the velocity profile was obtained in 88% of cases (44/49). The high values of the correlation coefficient observed with each measurement of the slope confirm the applicability of linear regression analysis to the pressure-velocity relationship. The slope of the pressure-velocity relationship was significantly lower in patients with coronary stenosis (1.7 +/- 0.7 cm/s/mmHg in normal vessels versus 0.7 +/- 0.3 cm/s/mmHg in stenotic arteries, p < 10(-4)), and, similarly, the pressure at zero flow was also reduced (36.9 +/- 16 mmHg versus 25.5 +/- 12 mmHg, p = 0.03). A statistically significant correlation was observed between the slope values and coronary flow reserve but no correlation was demonstrated between the slope and intraluminal surface area of angiographically normal coronary arteries or the slope and degree of stenosis of atheromatous vessels. The linear regression slope and the pressure at zero flow were lower when the pressure-velocity relationship was measured during long diastolic periods induced by the injection of adenosine. In addition, the curvilinear appearances of the pressure-velocity relationship observed during these long periods suggest that the linear regression model is not applicable throughout the whole range of pressures and velocities, especially for the lowest values.(ABSTRACT TRUNCATED AT 400 WORDS)