RESUMEN
This study evaluated the importance of the union between dentin collagen and three different adhesive materials. Sixty Class V restorations were prepared on the buccal and lingual surfaces of 30 recently extracted human premolars, with the cervical margins in dentin and the occlusal margins in enamel. These restorations were distributed to three groups of 20 cavities each based on the employed adhesive system used: Group A: Single Bond; Group B: Prime&Bond NT; Group C: One Coat Bond. Each group was subdivided according to dentin treatment: (1) manufacturers' adhesive protocol and (2) removal of the collagens fibers (total etch + sodium hypochlorite 5% for two minutes) + adhesive protocol. After the restorations were completed, the teeth were stored in saline solution (24 hours/37 degrees C), subjected to thermal cycling, washing and scoring according to dye penetration. Dye penetration was evaluated, with the numbers ranging from 0 (no infiltration) to 3 (greatest infiltration). When the dentin microleakage scores were compared in subgroups A1XA2, B1XB2 and C1XC2, the Mann-Whitney Test revealed significant differences between groups B and C (p < 5%). The better results were shown in subgroups B2 (p = 0.0345) and C2 (p = 0.0029). The results showed that the collagen fibrils were not necessary for adhesion, and their removal positively influenced the marginal sealing of Prime&Bond NT and One Coat Bond.