Mitral valve cusp separation on M mode echogram, the mitral valve opening artefact, and the onset of forward transmitral flow recorded by Doppler echocardiography have all been taken to mark the end of isovolumic relaxation, while its onset has been taken either as the aortic closure sound (A2) recorded phonocardiographically or the aortic closure artefact determined by Doppler technique. Possible differences in the measurement of the isovolumic relaxation time were studied when these landmarks were used in 44 healthy people, 14 patients with mitral stenosis, 21 patients with left ventricular hypertrophy, and 24 patients with dilated cardiomyopathy by recording M mode echograms of the mitral valve, and pulsed and continuous wave Doppler spectra of transmitral flow, with simultaneous electrocardiograms and phonocardiograms. A2 was effectively synchronous with the aortic artefact. However, when the onset of Doppler flow was regarded as the end of isovolumic relaxation, the interval was significantly longer than when mitral cusp separation on M mode echograms was used: by 25 (10) ms in healthy individuals, by 25 (15) ms in patients with left ventricular hypertrophy, and by 50 (35) ms in patients with dilated cardiomyopathy. In patients with mitral stenosis the interval was only 5 (5) ms longer. The mitral valve opening artefact consistently followed the onset of flow and corresponded much more closely to the E point on the M mode echogram. This shows that it occurred during the rapid filling period and well beyond isovolumic relaxation by any definition. Thus isovolumic relaxation time measured from A2 to the onset of transmitral flow or the mitral valve opening artefact differs from that derived from A2 to mitral valve cusp separation. These intervals cannot be used interchangeably to measure "isovolumic relaxation time".