Matrices of unsplinted attachment systems are generally reported to be the weak component of implant overdentures, often requiring frequent maintenance. Clinical wear results in reduced retention of the prosthesis, requiring activation or renewal of the matrix to restore the initial level of retention. The purpose of this retrospective study was to measure the wear of the matrix of a ball attachment after various periods of clinical wear. Seventy specimens of 3 groups of matrices of ball attachments that had been in use for mean periods of 12.3 months (1Y group, n=26), 39.0 months (3Y group, n=28) and 95.6 months (8Y group, n=16) were retrieved from 35 patients (2 specimens per patient) and measured on a coordinate measuring machine equipped with a touch trigger probe. Ten unused matrices were used as controls (CTRL group). The external and internal matrix diameters and deviations from circularity were measured. For the various time periods, the decreases in matrix thickness were calculated and compared with controls. Kruskal-Wallis 1-way ANOVA by ranks, followed by the Mann-Whitney post hoc tests, were conducted to test for differences in median values among groups (α =.05). For the internal upper diameter of the matrices tested, the Kruskal-Wallis and Mann-Whitney tests revealed significant differences for the 3 groups compared to the controls. For group 1Y, a significant difference (P<.001) of the internal upper diameter was found compared to the CTRL group. Compared to the controls, the nonparametric analyses for groups 3Y and 8Y showed significant differences for the internal upper diameter (P<.001) and deviations from circularity (P<.001). For groups 1Y, 3Y and 8Y, matrix thickness losses were 07, 47 and 70 μm, respectively. Within the limitations of this study, it was observed that one year of clinical wear had limited effect on the ball attachment matrices. Three to 8 years of clinical use resulted in a significant decrease of matrix thickness, especially at the tip of the retentive lamellae.