During the period 1968-1985, aortic valve replacement was performed in 871 patients, 617 males and 254 females aged 49.6 years in mean (15 to 75). Up to the end of 1988, 42 patients (4.8%) were reoperated on because of periprosthetic leakage. In primary surgery, annular calcification was found in 27 of these patients (64%), 17 patients (40%) had a bicuspid valve, and a positive culture from the excised valve was diagnosed in one of three patients with active native endocarditis. Two patients had Marfan's syndrome and two others postendocardial regurgitation. At primary surgery, mechanical prostheses were inserted in every case. Differences between the prosthetic models used could not be shown as regards to the occurrence of periprosthetic leakage. To implant the valve, interrupted everted U-mattress sutures with pledgets appeared to be the best method. Leakage occurred in 21 patients (50%) during their hospital stay and in 18 patients (43%) during the follow-up period of four months. Three others developed leakage two to four years from primary surgery. The main indication for reoperation was congestive heart failure alone or combined with heamolytic anaemia in 37 (88%) of cases. Four patients required the reoperation due to infectious dehiscence. Preoperatively, 34 patients (81%) were in NYHA (The New York Heart Association) Class III-IV. A new prosthetic valve was implanted in 23 cases, a composite graft once and in 18 cases refixation was performed. Two patients died in association with surgery, both due to the low output syndrome. During the follow-up time of 6.4 years, eight patients developed recurrent leakage. Four of them were found during their hospital stay and four others later. The role of prosthetic infection was remarkable in these eight cases; three patients with preoperative infectious dehiscence of the prosthesis had recurrence and one patient developed prosthetic endocarditis with leakage later. Three patients required more reoperations. At follow-up study, leakage was diagnosed in five of 28 living patients. Three of them were not significant and two moderate. 24 patients (86%) were in NYHA Class I-II.