The aim of the study was to assess the relationship between paroxysmal atrial fibrillation during acute myocardial infarction and the long-term prognosis of patients after acute myocardial infarction. The incidence of paroxysmal atrial fibrillation among 5803 consecutive hospitalized patients was 9.9% (557/5803). Incidence rose with increasing age (less than or equal to 59 years, 4.2%), (60-69 years, 10.5%), (greater than or equal to 70 years, 16.0%) and was slightly (but not significantly) higher in women (11.0%) than in men (9.6%). The presence of congestive heart failure and mean age represented two major discriminants between patients with paroxysmal atrial fibrillation (70% and 68.6 years) in comparison with their counterparts (35% and 62.3% years). Hospital mortality was significantly higher (25.5%) in patients with paroxysmal atrial fibrillation than in those without (16.2%). However, the effect of paroxysmal atrial fibrillation disappeared when other factors influencing the short term prognosis (i.e. heart failure) were taken into account by a multivariate logistic regression analysis. The covariate adjusted relative odds of in-hospital mortality then fell to 0.82. The 1- and 5-year mortality rates were 18.6% and 43.3% in patients with paroxysmal atrial fibrillation as compared to 8.2% and 25.4% (P less than 0.001), respectively, in patients free of paroxysmal atrial fibrillation. Using a proportional hazards analysis of mortality through the first quarter of 1988 (average follow-up time, 5.5 years) the net risk of dying among patients with paroxysmal atrial fibrillation complicating the acute myocardial infarction is estimated at 1.28 (90% confidence interval, 1.12-1.46) relative to counterparts free of the complication.(ABSTRACT TRUNCATED AT 250 WORDS)