Purpose: To evaluate the fatigue resistance of the bond between dentin and glass-infiltrated alumina ceramic, using different luting protocols. Materials and Methods: The null hypothesis is that the fatigue resistance varies with the luting strategy. Forty blocks of In-Ceram Alumina were prepared, and one surface of each block was abraded with 110-µm aluminum oxide particles. Then, the blocks were luted to flat dentin surfaces of 40 human third molars, using 4 different luting strategies (luting system [LS]/ceramic surface conditioning [CSC]) (n=10): (G1) [LS] RelyX-Unicem/[CSC] airborne abrasion with 110-µm Al2O3 particles; (G2) [LS] One-Step + Duo-Link (bis-GMA-based resin)/[CSC] etching with 4% hydrofluoric acid + silane agent; (G3) [LS] ED-Primer + Panavia F (MDP-based resin)/[CSC] Al2O3; (G4) [LS] Scotchbond1+RelyX-ARC (bis-GMA-based resin)/[CSC] chairside tribochemical silica coating (air abrasion with 30-µm SiOx particles + silane). After 24 h of water storage at 37°C, the specimens were subjected to 106 fatigue cycles in shear with a sinusoidal load (0 to 21 N, 8 Hz frequency, 37°C water). A fatigue survivor score was given, considering the number of the fatigue cycles until fracture. The failure modes of failed specimens were observed in a SEM. Results: G3 (score = 5.9, 1 failure) and G4 (score = 6, no failures) were statistically similar (p = 0.33) and had significantly higher fatigue resistance than G1 (score = 3.9, 5 failures) and G2 (score = 3.7, 6 failures) (p 0.03). SEM analysis of fractured specimens of G1 and G2 showed that almost all the failures were between ceramic and cement. Conclusion: The MDP-based resin cement + sandblasting with Al2O3 particles (G3) and bis-GMA-based resin cement + tribochemical silica coating (G4), both using the respective dentin bonding systems, were the best luting protocols for the alumina ceramic. The null hypothesis was confirmed. Parole chiave: fatigue resistance, alumina ceramic, luting systems, tribochemical silica coating