The choice of the ideal prosthesis for aortic valve replacement (AVR) in children is still controversial. Early degeneration of bioprostheses and the potential risks related to anticoagulation in the child have renewed the interest of many surgeons towards the Ross operation. This study concerns our 22-year experience with AVR in children. Forty-six children, aged 4 months to 16 years (mean 11.6 years), had AVR between April 1974 and December 1996. Preoperative diagnosis revealed aortic regurgitation (AR) in 25 cases, aortic stenosis (AS) in ten, combined AS and AR in nine and LVOTO in two patients. Of the 46 patients, 26 had 37 previous procedures. Isolated AVR was performed in 19 cases, 27 children underwent 36 concomitant intracardiac procedures. Mechanical prostheses were implanted in 30 children, bioprostheses in eight, aortic homografts in three. Five patients underwent a Ross procedure. There was one hospital death in the latter group (2.1%). Six of seven late deaths occurred in patients who underwent complex intracardiac procedures (15.2%). Reoperation rate was 19.5% (n = 10), differentiating 16.6% for mechanical (5/30 patients) and 50% (4/8 patients) for bioprostheses. The mean follow-up period was 8.01 years, ranging from 0.45 to 21.66 years (304.04 patient-years). There was one hemorrhagic event (2.1%) due to anticoagulation, thrombosis of the mechanical valve occurred in two patients (4.2%). AVR can be performed with acceptable mortality rate and good long-term results in children. We perform the Ross operation only in selected cases. According to our experience, mechanical prostheses show excellent performances in children with a low incidence of complications related to anticoagulation.