Objectives: This study tested the hypothesis that the size of repair sites on amalgam restorations with mechanical undercuts on the repair surface would affect interfacial strength using a 3-point bending test. Method and Materials: Six groups of 12 rectangular bars each were fabricated from an admix alloy and a spherical alloy. Each specimen was cut into 2 equal-sized end sections, and 1 smaller midsection, which was discarded. A medium-coarse diamond disk was used to roughen and flatten the surface for repair. The 2 end sections were fit back into the original mold. The empty space in the middle was approximately 1, 1.5, and 2 mm, mimicking the size of repair sites to be filled with amalgam. There were 2 groups for each repair size; 1 group had rounded undercuts prepared on the surface for repair with a 1/2 round bur, and the other group received no additional treatment. Condensation was carried out with condensers having a dimension slightly less than the width of the repair defect. After 7-day storage, a 3-point bending test was used to determine the repair strengths of the specimens. Results: The strength values ranged from 33.9 to 75.6 MPa, which is comparable to published data. ANOVA showed that the mean strength values were higher with spherical amalgam (P .0001), were lower with undercut (P = .0148), and increased as the width of repair site decreased (P .0001). Conclusion: Reducing the dimension of the repair site improved repair strength, but rounded undercuts reduced the repair strength values slightly. Keywords: admix, amalgam, condensation, defect, flexure strength, repair, spherical