Background Previous studies have shown that investigation by tilt testing is very appropriate in paediatrics, but the characteristics of children and adolescents who are at high risk of recurrent syncope, once the diagnosis is established, remain unclear. This study was set up to analyse the risk factors attributed to syncope recurrence in paediatric patients. Methods One hundred and one children and adolescents aged 7 to 18 years, undergoing a tilt test for recurrent syncope, were studied. They were subsequently followed-up in clinic visits with a final interview at the clinic or by telephone at the end of the follow-up period. Results A head-up tilt test elicited syncope or pre-syncope in 67 children. The positive responses included vasovagal syncope in 58 patients and psychogenic syncope in nine patients. Gender, age, number of pre-tilt test syncopal episodes or duration of symptoms made no difference to children with positive or negative tilt test results. Following the tilt test, 43 of 67 children with a positive tilt test were treated empirically. No treatment was prescribed for the remaining 24 with a positive test, or for those with a negative tilt test. There were no differences between treated and untreated children with respect to the number of pre-tilt test syncopes, duration of symptoms and duration of follow-up. Follow-up data were available in 97 children. During a mean follow-up of 46±28 months, syncope recurred in 31 children (32%). The recurrence rate was similar between positive and negative tilt test groups (22/66 vs 9/31, respectively;P=ns), as well as between treated and untreated children (14/43 vs 8/23, respectively;P=ns). When comparing syncope-free children at follow-up in a univariate analysis, children with recurrent syncope reported a greater number of historical syncopal spells (7±8 vs 3±3,P =0·01). In addition linear correlation (r=0·6, 95% CI 0·47 to 0·72,P Conclusions These findings suggest that the risk of syncope recurrence for children and adolescents with such a history is not correlated to the tilt test result or prophylactic treatment. The number of historical syncopal spells is, however, predictive.