Background The long-term results of mitral valve replacement (MVR; n = 631) with a bileaflet mechanical prosthesis or a Carpentier-Edwards Perimount bioprosthesis were evaluated in Japanese patients of different age groups. Methods A total of 507 bileaflet mechanical prostheses and 124 bioprostheses have been implanted since 1982 at our institution. Follow-up was completed for 6,598 patient-years in 98.4% of the cases. Results Among the patients 70 years of age and older, the rate of freedom from valve-related death and valve-related morbidity at 10 years after surgery were significantly better in the bioprostheses group (93.3% ± 6.4% and 83.7% ± 8.7%, respectively; n = 35) than in the mechanical prostheses group (71.1% ± 8.0% and 60.9% ± 8.9%, respectively; n = 82), and neither structural valve deterioration (SVD) nor resulting re-MVR were observed for bioprostheses. In contrast, among the patients 64 years and younger, no significant differences were observed in long-term survival between the mechanical prostheses group (n = 347) and the bioprostheses group (n = 76), while significantly lower rates of freedom from SVD and re-MVR were observed in the bioprostheses group compared with those obtained in the mechanical prostheses group. As for the controversial intermediate-age group of 65 to 69 years, the general tendencies were similar to those observed in the group 64 years and younger. Conclusions Based on our comparative evaluation, bioprostheses should be chosen for MVR in patients 70 years of age and older, whereas mechanical prostheses were better in the patients 64 years of age and younger. The use of bioprostheses in Japanese patients 65 to 69 years of age is not preferable for preventing SVD and subsequent re-MVR.