Background: Doppler tissue imaging (DTI) is reported to be an easy, fast, and reliable method to estimate the atrial mechanical activation time (aMAT), which is known to be well correlated with total P-wave duration measured by signal-average electrocardiogram (SAECG). However, it remains to be determined whether these parameters are matched or not in patients with atrial fibrillation (AF). Methods and Results: Forty one paroxysmal AF (PAF) group (mean age, 54.2짹13.5 years) and age/sex-matched controls group (n=131, mean age, 53.3짹7.6 years) were investigated off antiarrhythmic drugs. We determined aMAT as the time interval from initiation of the P-wave (lead II) deflection until the peak of the local lateral left atrial (LA) DTI signal in apical 4-chamber view. Atrial electrical activation time (aEAT) was defined as total P-wave duration using SAECG. There was strong correlation between aEAT and aMAT in the control group (R=0.488, p<0.01) and marginally significant correlation was observed in the PAF group (R=0.309, p=0.05). When the PAF group was divided into two groups by its frequency of episode (��� or < 1 attack per week), good correlation was still preserved in the infrequent PAF group (n=29, R=0.486, p<0.01), whereas this correlation was totally lost in the frequent PAF group (n=12, R=0.035, p=0.91). Conclusions: Compared to controls, the correlation between aEAT and aMAT was significantly attenuated according to the frequency of AF episodes. This mechano-electrical mismatch might be caused by inhomogeneous substrate of the atria, resulted in the development of AF.