Objective and methodsTo determine whether hypertensive patients at risk for paroxysmal atrial fibrillation (AF) could be detected while in sinus rhythm, a computer-based 12-lead surface electrocardiogram was recorded in 50 hypertensive patients with history of paroxysmal AF (group A) and in 60 hypertensive patients without history of AF (group B). The maximum P-wave duration (Pmaximum), the minimum P-wave duration (Pminimum), P-wave dispersion (Pdispersion= Pmaximum− Pminimum), adjusted P-wave dispersion (APdispersion= Pdispersion/square root of the number of measurable leads), mean P-wave duration (mean P) and the standard deviation of the P-wave duration in all measured leads (SDP) were calculated.ResultsPdispersion, APdispersionand SDP were significantly higher in group A than in group B (Pdispersion, 52 ± 19 versus 41 ± 15 ms,P< 0.001; APdispersion, 15.2 ± 5.5 versus 11.9 ± 4.6 ms,P< 0.001; SDP, 16 ± 5 versus 13 ± 5 ms,P< 0.001). Pminimum, mean P and left ventricle ejection fraction (LVEF) were significantly lower in group A than in group B (Pminimum, 79 ± 18 versus 91 ± 13 ms,P< 0.001; mean P, 108 ± 18 versus 116 ± 13 ms,P= 0.005; LVEF, 64 ± 5 versus 69 ± 8%,P< 0.001). Pminimum, Pdispersion, mean P, SDP, APdispersionand LVEF were found to be significant univariate predictors of paroxysmal AF, whereas only Pminimum(P< 0.001) remained a significant independent predictor of paroxysmal AF in the multivariate analysis.ConclusionHypertensive patients at risk for paroxysmal AF could be detected while in sinus rhythm by computer-assisted electrocardiographic P-wave analysis.