Impaired long-axis motion is a sensitive marker of systolic myocardial dysfunction, but no data are available that relate long-axis changes in systole with those in diastole, particularly in subjects with diastolic dysfunction and a 'normal' left ventricular (LV) ejection fraction. A total of 311 subjects (including 105 normal healthy volunteers) aged 20-89 years with variable degrees of systolic function (LV ejection fraction range 0.15-0.84) and diastolic function were studied using tissue Doppler echocardiography and M-mode echocardiography to determine mean mitral annular amplitude and peak velocity in systole and early and late diastole. The LV systolic mitral annular amplitude (S(LAX), where LAX is long-axis amplitude) and peak velocity (S(m)) correlated well with the respective early diastolic components (E(LAX) and E(m)) and late diastolic (atrial) components (A(LAX) and A(m)). A non-linear equation fitted better than a linear relationship (non-linear model: S(LAX) against E(LAX), r(2)=0.67; S(m) against E(m), r(2)=0.60; S(LAX) against A(LAX) and S(m) against A(m), r(2)=0.42). After adjusting for age, sex and heart rate, linear relationships of early diastolic (E(LAX), r(2)=0.70; E(m), r(2)=0.60) and late diastolic (A(LAX), r(2)=0.61; A(m), r(2)=0.64) long-axis amplitudes and velocities with the respective values for S(LAX) and S(m) were found, even in those subjects with apparently 'isolated' diastolic dysfunction. Long-axis changes in systole or diastole did not correlate with Doppler mitral velocities. We conclude that ventricular long-axis changes in early diastole are closely related to systolic function, even in subjects with diastolic dysfunction. 'Pure' or isolated diastolic dysfunction is uncommon.