Objective: To compare marginal quality of all-ceramic restorations varying the etching (phosphoric acid-PA versus self-etching primer-SE), activation mode of the adhesive (light-cure-LC versus dual-cure-DC), and substrate (enamel or dentin). Methods: Forty porcelain inlays were cemented to standardized MOD cavities of extracted human molars combining the etchings and activation modes (n=10) and following manufacture's instructions. In each tooth, one of the gingival walls was in enamel and the other in dentin. The adhesive used was Optibond Solo Plus (Kerr) in its different versions (see Table). The dual-cure resin cement was Nexus (Kerr). After thermal cycling, marginal gaps were analyzed using SEM of epoxy resin replicas in order to determine the percentage length of the debonded margins relative to the gingival wall periphery. Data were analyzed by ANOVA and Tukey's test (alpha=5%). Results: Mean gap percentages of different groups are in the Table. For the groups etched with PA, mean values obtained in enamel were significantly lower compared to those obtained in dentin, regardless of the activation mode. All groups etched with SE presented similar gap percentages, regardless of the activation mode and substrate. The comparison PA versus SE shows that, in enamel, the first resulted in lower gap formation, regardless of the activation mode. No statistical difference was found between mean values of different activation modes, considering the same type of etching and substrate. Conclusion: PA led to less gap formation than SE only in enamel; and the activation mode of the adhesive did not affect gap formation in bonded porcelain restorations.

Etching	Activation Mode	Substrate
		Enamel	Dentin
PA	LC	18±5 b	32±7 a
	DC	20±8 b	35±8 a
SE	LC	28±10 a	31±10 a
	DC	32±9 a	38±9 a