The impact of anxiety neurosis on the diagnosis and treatment of patients with unexplained syncope (S) was assessed in 178 patients (91 women and 87 men) with an average age of 36.5 +/- 20 years, presenting with 10.7 +/- 24 episodes of S). None had evidence of underlying cardiac disease apart from 7 patients with mild hypertension. All patients underwent a tilt test (TT) at 60 degrees for 45 minutes. A bolus of isoproterenol was injected intravenously in subjects with negative TT. After the test, the patients were classified according to the presence (n = 38) or absence (n = 140) of anxiety neurosis based on the DSM III-R diagnostic. The TT was positive in 76 patients, 9 of whom had a cardioinhibitory reaction with prolonged asystole. Patients with anxiety had more episodes of S (24 +/- 43 versus 7 +/- 13; p = 0.001), a shorter interval between S (11.5 +/- 23 months versus 12.5 +/- 20 months, p = 0.02) but more negative TT (27/38 versus 75/140; p = 0.05). One hundred and sixty-eight patients were followed up : 10 were lost to follow-up. Preventive treatment was undertaken in 59 patients who were representative of the whole group with respect to age (30 +/- 18 years 39 +/- 21 years : p = 0.004). After an average follow-up of 24.5 +/- 15 months, 26 patients (15%) experienced a recurrence of S. The recurrence rate was identical in patients with positive and negative TT and in treated and untreated cases. On the other hand, recurrence was higher in those with anxiety (12/25 versus 14/117; p = 0.001) who also had less improvement of symptoms (12/15 versus 74/120; p = 0.001). The "anxiety" variable was therefore identified as being the only predictive factor for recurrence of syncope. The authors conclude that in patients referred for investigation of unexplained syncope, some suffer from anxiety neurosis, in whom the TT is usually negative, and have a higher risk of recurrence. They justify a specific therapeutic management.