Purpose: To evaluate the influence of different surface treatments on roughness and bond strength of composite repairs. Materials and Methods: 120 truncated conical specimens were prepared with composite Grandio SO (VOCO) and submitted to 5000 thermal cycles. Specimens were divided into 12 groups (n = 10) regarding the surface treatments: negative control (NC), without treatment; medium-grit diamond bur (MGD); coarse-grit diamond bur (CGD); conventional carbide bur (ConC); crosscut carbide bur (CutC); chemical vapor deposition diamond bur (CVD); sandblasting with aluminum oxide (AlO); Er:YAG laser 200 mJ/10 Hz (Er200); Er:YAG laser 60 mJ/10 Hz (Er50); Nd:YAG laser 120 mJ/15 Hz (Nd120); Nd:YAG laser 60 mJ/ 15Hz (Nd60); air abrasion with 110-μm silica modified aluminum oxide (Rocatec Plus-3M) (SIL). After the surface treatments, the surface roughness (Ra) was measured using a profilometer, and then the adhesive system Admira Bond (VOCO) was applied. Another truncated conical restoration was built up with the same composite over the bonded area of each specimen. In order to evaluate the cohesive strength, double-cone specimens were made and considered as a control group (CoheC). The specimens were submitted to tensile bond strength testing and the obtained data (MPa) were evaluated by one-way ANOVA, Tukey's and correlation tests. Results: ANOVA showed significant differences among experimental groups for roughness and adhesive strength (p 0.00). The roughness values (Ra) were: NC (0.21 ± 0.19)c; ConC (0.30 ± 0.08)c; CutC (0.50 ± 0.22)cd; CVD (0.74 ± 0.14)bd; MGD (0.89 ± 0.39)ab; Er50 (0.89 ± 0.14)ab; AlO (0.90 ± 0.07)ab; Nd60 (0.94 ± 0.33ab; SIL (0.98 ± 0.07)ab; Nd120 (1.10 ± 0.19)a; CGD (1.10 ± 0.32)a; Er200 (1.12 ± 0.21)a. The results of the tensile bond strength test in MPa were: CGD (11.58 ± 3.03)a; MGD (12.66 ± 3.82)ab; NC (13.51 ± 3.95ab; Nd120 (14.11 ± 5.95)ab; ConC (14.73 ± 6.12)ab; Er200 (15.51 ± 1.45)abc; CVD (15.61 ± 5.00abc; Er50 (16.44 ± 2.75) abc; CutC (16.79 ± 2.98)abc; Nd60 (17.72 ± 2.45)abcd; AlO (18.33 ± 3.19)bcd; SIL (21.13 ± 4.48cd; CoheC (23.50 ± 5.81)d. The groups followed by the same letters were not statistically significantly different (Tukey's test). No correlation was found between bond strength and roughness (r = 0.007). Conclusion: Air abrasion with silica coating (Rocatec) was the only method which resulted in significantly higher bond strength in relation to the negative control group. The increase in laser energy produced a rougher surface, but reduced the bond strength. Keywords: composite resin, composite repair, roughness, bond strength