To investigate factors during isovolumic relaxation that determine Doppler filling patterns in patients with left ventricular disease, and thus to identify the underlying mechanisms. 85 patients (50 ischaemic heart disease, 35 left ventricular hypertrophy due to aortic stenosis) and 26 controls were studied with Doppler and M mode echocardiography and phonocardiography. 16 patients underwent two studies on separate occasions, to find whether changes in isovolumic relaxation time were reflected by a change in the Doppler A/E ratio. A tertiary cardiac referral centre. Patients referred for assessment of coronary artery disease or aortic stenosis with left ventricular hypertrophy. Doppler filling velocities during early (E wave) and late (A wave) diastole and the A/E ratio, acceleration of the E wave, digitised M mode indices of incoordinate relaxation (change in cavity dimension before mitral valve opening and time from minimum dimension to mitral valve opening), isovolumic relaxation time, M mode measures of diastolic function after mitral valve opening (peak rate of posterior wall thinning and peak rate of dimension increase), and left ventricular end diastolic pressure. A/E correlated with age in normal subjects (r = 0.74), to a lesser extent in left ventricular hypertrophy (r = 0.41), but not significantly in ischaemic heart disease. In all patients, isovolumic relaxation time was significantly and negatively correlated with the acceleration of the E wave, showing its fundamental relation to the force responsible for early diastolic filling (r = -0.71 for left ventricular hypertrophy, and -0.74 for ischaemic heart disease, p value < 0.01). In left ventricular hypertrophy and those ischaemic patients without left ventricular dilatation A/E was correlated both with isovolumic relaxation time (r = 0.68 and 0.60 respectively), and with incoordinate relaxation (r = 0.65 and 0.61). In those ischaemic patients with left ventricular dilatation, the influence of incoordination was lost and isovolumic relaxation time became the dominant influence upon A/E (r = 0.82). Weak correlations of end diastolic pressure and RR interval with A/E, became insignificant once isovolumic relaxation time had been taken into account. Isovolumic relaxation time and incoordination together accounted for over 50% of the variance in the A/E ratio in our patients. Isovolumic relaxation time and the A/E ratio were linearly related. Patients with a short isovolumic relaxation time had evidence of considerable diastolic abnormalities, despite a normal Doppler A/E ratio. In the 16 patients who had two echocardiographic studies, changes in the duration of isovolumic relaxation were accompanied by a change in the Doppler A/E ratio. The relation between these two variables, derived from the group as a whole was similar. The main factors influencing the A/E ratio in patients with left ventricular disease are two distinct properties of isovolumic relaxation--namely the duration and the extent of incoordinate wall motion. Filling pressure and RR interval are not significant independent determinants, but act only through an effect upon isovolumic relaxation time. Age is an important influence in normal people, but this effect is attenuated in left ventricular hypertrophy and lost in ischaemic ventricular disease.