Multivariate analysis using 70 variables in 200 patients who suffered from ventricular tachycardia or ventricular fibrillation after myocardial infarction detected eleven variables that were associated with an increased risk of sudden arrhythmic death and cardiac death during a mean follow-up period of 2 years. Four of the 11 variables came from the patient's clinical history: (1) cardiac arrest at the time of the first spontaneous episode of arrhythmia, (2) New York Heart Association functional class for dyspnoea = III, (3) ventricular tachycardia or ventricular fibrillation occurring early (after 3 days and within 2 months) after myocardial infarction, (4) multiple myocardial infarctions before the first episode of ventricular tachyarrhythmia. Total mortality, incidence of sudden arrhythmic death and of non-sudden cardiac death increased with an increasing number (zero, one, two, three, four) of variables seen in individual patients. Patients with zero or one variable had an incidence of sudden death of 2.8% and a 4.2% incidence of non-sudden cardiac death at 26 months, while patients with more than two variables had a 13.5% and a 20.3% incidence respectively of sudden and non-sudden cardiac death. The strongest predictor of sudden death was the occurrence of cardiac arrest during the first spontaneous episode of ventricular arrhythmia. The strongest predictor of non-sudden cardiac death was the New York Heart Association functional class. The use of the four variables to stratify risk revealed seven subgroups of patients with incidences of sudden death ranging from 0 to 28%.(ABSTRACT TRUNCATED AT 250 WORDS)