Villani GQ, Piepoli M, Cripps T, Rosi A, Gazzola U. General Hospital, Piacenza, Italy. Pacing Clin Electrophysiol 1994 17(6):1118-23. High gain, signal-averaged ECGs using conventional surface lead technique and a transesophageal lead technique were performed in 45 idiopathic paroxysmal atrial fibrillation patients and in 33 normal controls. Both techniques showed increased P wave duration in patients compared with the controls (P < 0.001), but higher P wave amplitudes were obtained using the transesophageal technique compared with surface leads (patients: 169.8 +/- 81.7 microV vs 15.8 +/- 7.3 microV; P < 0.0005; controls: 163.5 +/- 22.1 microV vs 18.5 +/- 5.2 microV; P < 0.0005). The signal-averaged transesophageal lead, but not the surface recordings, identified the presence of atrial late potentials evidenced by lower root mean square voltages in the terminal portion of the P wave: in last 10 seconds, 4.4 +/- 1.3 microV versus 8.5 +/- 3.0 microV; P < 0.001; in last 20 seconds, 7.0 +/- 2.3 microV versus 16.0 +/- 7.9 microV; P < 0.001; in last 30 seconds, 12.5 +/- 5.3 microV versus 23.8 +/- 12.8 microV; P or = 110 msec had 85% sensitivity, 100% specificity, and 100% positive predictive value in identifying the patients; the combined criteria P wave duration > or = 110 msec and root mean square for the last 10 msec < or = 6.5 showed 80% sensitivity, 100% specificity, and 100% predictive value. The signal-averaged transesophageal lead produces a higher amplitude signal, which reveals fractionation of atrial activation in atrial fibrillation and allows identification of individuals predisposed to this arrhythmia.