QRS duration is one of the most important factors analysed by signal-averaged electrocardiography. The purpose of the study was to compare three methods of QRS duration measurement (conventional 12 lead ECG, Frank vectorcardiogram [VCG] and signal-averaged electrocardiography [SA ECG]) and to look for the factors associated with changes in QRS duration. The recordings were made at the same time in 113 consecutive patients with a PC-based system ECG and VCG (Cardionics) and high resolution ECG (Cardionics) based on methods described by Simson. Patients with bundle branch block were excluded. All patients had presented a myocardial infarction and were studied either for spontaneous ventricular arrhythmias or systematically, early after infarction (3 to 6 weeks) in 45 patients (group I) or late after infarction (> 9 months) in 68 patients (group II). 21 patients of group I had inducible sustained monomorphic ventricular tachycardia (VT) (group I) and 24 did not (group Ib). 36 patients of group II had inducible VT (group IIa) and 32 did not (group IIb). The comparison of the 3 methods for the evaluation of QRS duration shows that the QRS duration was related to 2 factors, the inducibility of VT, and the age of the infarction: 1. QRS duration was prolonged in patients with inducible sustained VT in groups I and II. 2. QRS duration was longer in patients of group I than in group II. A value > 110 ms in group I was pathological, while a value > 120 ms was required ingroup II to be considered as pathological.(ABSTRACT TRUNCATED AT 250 WORDS)