Pulsed tissue Doppler imaging was performed to measure systolic left ventricular (LV) posterior wall motion velocity along the long and short axes and to evaluate the relationships between the systolic variables and the maximum first derivative (peak dP/dt) of the LV pressure curve and the 2 preceding R-R intervals in 39 patients with atrial fibrillation (AF). The study group consisted of 22 patients with AF only, 17 patients with dilated or ischemic cardiomyopathy and AF, and 25 healthy control subjects in sinus rhythm. The systolic component of the LV posterior wall motion velocity was divided into the first (Sw1) and second (Sw2 ) systolic waves. The peak Sw1 along the long axis was greater than either that along the short axis or the peak Sw2s along the long and short axes in the control and AF-only groups. The peak Sw1 along the long axis was lower in the AF-only group than in the control group, and those along the short and long axes were lower in the dilated AF group than in the other groups. The peak Sw1 almost coincided with the peak dP/dt. The peak Sw1 along the long axis correlated closely with the peak dP/dt, and the ratio of the preceding R-R interval to the interval before the preceding ("prepreceding") R-R interval in both AF groups, particularly in the dilated AF group, and the slopes of their relationships were steeper in the dilated AF group than in the AF-only group. The peak Sw2 along the short axis was greater than that along the long axis in the control and AF-only groups. The peak Sw2 along the long axis was lower in the AF-only group than in the control group, and those along the short and long axes were lower in the dilated AF group than in the other groups. The peak dP/dt was lower and the LV end-diastolic pressure was higher in the dilated AF group than in the other groups. In conclusion, peak Sw1 along the long axis is useful for the evaluation of isovolumic myocardial LV contractility, and the interval-force relation and the Frank-Starling mechanism are important factors of beat-to-beat variability in systolic LV function in patients with AF.