OBJECTIVE--To assess the prevalence of ventricular late potentials and ventricular tachycardia in hypertensive subjects with left ventricular hypertrophy and to study their relation to clinical characteristics. SETTING--Teaching and general hospital in Padua. METHODS--107 hypertensive subjects with echocardiographic signs of left ventricular hypertrophy were studied with signal averaged electrocardiography and 24 hour Holter monitoring. Signal averaged electrocardiogram analysis was performed with high pass filters of 25 Hz, 40 Hz, and 80 Hz. Ventricular late potentials were considered to be present if at least two determinants of the signal averaged electrocardiogram were abnormal in one of the three filters. 70 normotensive subjects served as age matched controls. RESULTS--25% (27) of the hypertensive subjects and 6% (four) of the controls showed late potentials on signal averaged electrocardiography (P < 0.0001). The hypertensive subjects with late potentials had a higher prevalence of ventricular tachycardia (33%, 9/27) than those without late potentials (13%, 10/80; P = 0.035). Twenty nine per cent (31/107) of the hypertensive subjects had an inversion of the early to atrial filling velocity (E/A ratio < 1) on Doppler analysis of transmitral flow. Within this group the percentage of subjects with late potentials (55%, 17/31) and ventricular tachycardia (42%, 13/31) was much greater than that within the group of subjects without an inverted E/A ratio (13%, 10/76 (P < 0.0001) and 12%, 9/76 (P = 0.001) respectively). In a multivariate analysis only the E/A ratio was related to the presence or absence of either late potentials (P = 0.0001) or ventricular tachycardia (P = 0.0008). Both late potentials and ventricular tachycardia were unrelated to left ventricular mass, geometry, and systolic performance. CONCLUSIONS--A relation was found between the occurrence of ventricular tachycardia and the presence of late potentials in hypertensive subjects with left ventricular hypertrophy. Impaired left ventricular filling was the main marker for the arrhythmogenic substrate present in this disease.