The aim is the analysis of the P wave on the signal averaged ECG in 31 pts: 12 control pts (6 M, 6 W, 40 +/- 10 y) 12 HTA (9 M, 3 W, 60 +/- 7 y), 7 pts (5 M, 2W, 48 +/- 7 y) with sustained paroxystic atrial fibrillation (AF) without organic heart disease, without antiarrhythmic drugs. We measured the filtered P wave duration (Ad), the integral of Ad, the root mean square voltage of Ad for the last 10, 20, 30, 40, 60 msec and the duration of P wave on the ECG in lead II (P II) and the echocardiographic dimensions of the atria (LAd). HTA Ad (132 +/- 12 msec)* et > control Ad (116 +/- 10 msec) HTA LAd (38 +/- 3 mm) et > control LAd (31 +/- 0.7 mm) HTA PII (120 +/- 1.5 mm)* et > control PII (88 +/- 10 mm). The difference between HTA Ad (132 +/- 12 msec) and AF Ad (129 +/- 7 msec) is not significant. The linear regression tests don't show correlation between P II and Ad and between LAd and Ad in HTA group. There is a correlation between Ad and LAF in AF group (r = 0.83, p 0.02). HTA RMS 2o (2.2 + 0.6 microV), control RMS 2o (3.9 + 1.8 V) but HTA RMS 2o and AF RMS 2o (2.4 +/- 0.6 microV) are not significantly different and are not correlated with LAd and PII. A long duration of P filtered P wave and a low RMS 2o observed in HTA group and AF group would be a criteria of atrial vulnerability. p < 0.05.