To evaluate clinical usefulness of sound spectral analysis in the early detection of prosthetic thrombosis, we analyzed specific patterns of thrombus formation in Björk-Shiley tilting standard disc prostheses in relation to the sound spectral analysis of their click sounds. Among a total of 365 Björk-Shiley tilting standard disc prostheses, nine became thrombotic. These consisted of seven prostheses in the mitral position and two in the aortic position, and the affected valves were replaced in all cases. The sound spectral analysis system includes a wide-frequency microphone and a sound spectral analyzer. To quantify the amplitude of the metallic clicks that are specific to the prosthetic valve sounds, we used a normalized maximal frequency as a diagnostic parameter for valve thrombosis. Thrombus formation was localized at the minor strut in two cases, and these demonstrated abnormally low normalized maximal frequency values only on opening clicks. In four cases thrombus formations were noted at both minor and major struts, and these showed abnormally low normalized maximal frequency values on both opening and closing clicks. In two cases with pannus formation, no abnormalities were found by sound spectral analysis. In one case, which had demonstrated abnormally decreased normalized maximal frequency values of both opening and closing clicks, the normalized maximal frequency values were normalized after thrombolytic therapy, suggesting resolved thrombosis. Reoperation, which was performed for concomitant complication of perivalvular leak, revealed no noticeable thrombus formation. The thrombi were seen on the minor strut during surgical procedures in all cases in which valve thrombosis was indicated by the sound spectral analysis. Therefore, the minor strut is considered to be the prevalent area of thrombus formation, and diagnosis could be made after observation of abnormally low normalized maximal frequency on opening clicks. We found the sound spectral analysis to be an extremely useful diagnostic tool for early detection of thrombosis in the Björk-Shiley tilting standard disc prostheses, especially because the opening clicks of the tilting disc valve were too low in amplitude for auscultation to detect the existence of mild abnormality.