Abstract: Objective: To compare the fracture resistance of ceramic veneers and composite resin veneers with or without dental preparation. Methods: Forty freshly extracted mandibular premolars were selected and randomly assigned into four groups (n=10): NPR =no dental preparation and direct veneer with 0.2 mm thick composite resin; NPC = no dental preparation and 0.2 mm thick lithium disilicate ceramic veneer (IPS e.max Press, Ivoclar Vivadent); P2C = tooth preparation of 0.2 mm and 0.2 mm thick ceramic veneer (IPS e.max Press); and P5C = tooth preparation of 0.5 mm and 0.5 mm-thick ceramic veneer (IPS e.max Press). In all groups, the restorations covered 1 mm of the occlusal surface of the buccal cusp, and the thickness of this area was the same as the buccal area (0.2 mm or 0.5 mm). After the luting procedure, all groups were thermocycled (10,000 cycles, 5 ℃~55 ℃) and subjected to fracture resistance test under compression (Instron 4 444). The mode of failure analysis was performed under a ×10 magnification. Results: The mean fracture resistance (men ± standard deviation) was NPR =690.33±233, NPC=790.52±408, P2C=1 131.34±341, and P5C=983.56±202. There were significant differences of the fracture resistance values between all groups (P<0.05). The mean values of NPR and NPC groups were significantly lower than that of P2C. However, P5C presented intermediate values and had no significant difference from the other groups. The mode of failure for all groups was mixed (60%), cohesive failures (20%), root failures (15%), and adhesive failures (5%). Conclusion: Minimally invasive tooth preparation (0.2 mm) allowed achieving higher fracture resistance in premolars restored with lithium disilicate ceramic veneers. Attention should be given to the 0.5 mm preparation since fractures could happen at this situation.

Key words: ceramic veneers, fracture resistance, lithium disilicate, mode of failure, resin composite