Ectopic atrial tachycardia (EAT) is usually considered as benign and easy to treat. The natural history of the disease, however, has not yet been clarified. The purpose of the study was to analyse its spontaneous evolution in a cohort of EAT patients and to define a predictive model of remission based on several factors. Between 1973 to 1989, 46 patients (25 male, 21 female), aged 38 +/- 18 years, entered the study. Clinically EAT was paroxysmal in 23 patients, permanent in 12 and repetitive in 11; six patients were asymptomatic. Thirty-five complained of palpitations; dyspnoea, dizziness and syncope were also reported less frequently. All patients underwent an electrophysiological study to clarify the mechanism of the arrhythmia and to localize its site of origin. In 15 patients no heart disease was documented. Five patients underwent surgery and were excluded from subsequent analysis. Seven patients were discharged without antiarrhythmic treatment. We defined remission as the absence of recurrence of EAT within 6 months from withdrawal of therapy. Logistic regression was applied to identify potential predictors of remission. Seven clinical and electrophysiological covariates were entered in the model; univariate and multivariate tests were performed, using the GLIM3 statistical package. During a follow-up period of 5.1 +/- 4.5 years, 14 instances of remission (34%) were observed in 5/22 patients with paroxysmal EAT, 4/8 patients with permanent EAT and 5/11 patients with repetitive EAT. Mean age of patients with remission was 25 +/- 14 years vs 45 +/- 15 years in the group without remission. No covariate had an independent predictive value.(ABSTRACT TRUNCATED AT 250 WORDS)