The aims of this study were to evaluate the shear bond strength (SBS) of metal and ceramic brackets bonded to two different all-ceramic crowns, IPS Empress 2 and In-Ceram Alumina, to compare the SBS between hydrofluoric acid (HFA), phosphoric acid etched, and sandblasted, non-etched all-ceramic surfaces. Ninety-six all-ceramic crowns were fabricated resembling a maxillary left first premolar. The crowns were divided into eight groups: (1) metal brackets bonded to sandblasted 9.6 per cent HFA-etched IPS Empress 2 crowns; (2) metal brackets bonded to sandblasted 9.6 per cent HFA-etched In-Ceram crowns; (3) ceramic brackets bonded to sandblasted 9.6 per cent HFA-etched IPS Empress 2 crowns; (4) ceramic brackets bonded to sandblasted 9.6 per cent HFA-etched In-Ceram crowns; (5) metal brackets bonded to sandblasted 37 per cent phosphoric acid-etched IPS Empress 2 crowns; (6) metal brackets bonded to sandblasted 37 per cent phosphoric acid-etched In-Ceram crowns; (7) metal brackets bonded to sandblasted, non-etched IPS Empress 2 crowns; and (8) metal brackets bonded to sandblasted, non-etched In-Ceram crowns. Metal and ceramic orthodontic brackets were bonded using a conventional light polymerizing adhesive resin. An Instron universal testing machine was used to determine the SBS at a crosshead speed of 0.1 mm/minute. Comparison between groups was performed using a univariate general linear model and chi-squared tests. The highest mean SBS was found in group 3 (120.15 +/- 45.05 N) and the lowest in group 8 (57.86 +/- 26.20 N). Of all the variables studied, surface treatment was the only factor that significantly affected SBS (P < 0.001). Acid etch application to sandblasted surfaces significantly increased the SBS in groups 1, 2, 5, and 6. The SBS of metal brackets debonded from groups 1, 3, and 5 were not significantly different from those of groups 2, 4, and 6. All debonded metal brackets revealed a similar pattern of bond failure at the adhesive-restorative interface. However, ceramic brackets had a significantly different adhesive failure pattern with dominant failure at the adhesive-bracket interface. Ceramic fractures after bracket removal were found more often in groups 1-4. No significant difference in ceramic fracture was observed between the IPS Empress 2 and In-Ceram groups.