RÉSUMÉ
OBJECTIVE: To evaluate the gingival marginal seal in class II composite restorations using different restorative techniques. MATERIAL AND METHODS: Class II box cavities were prepared in both proximal faces of 32 sound human third molars with gingival margins located in either enamel or dentin/cementum. Restorations were performed as follows: G1 (control): composite, conventional light curing technique; G2: composite, soft-start technique; G3: amalgam/composite association (amalcomp); and G4: resin-modified glass ionomer cement/composite, open sandwich technique. The restored specimens were thermocycled. Epoxy resin replicas were made and coated for scanning electron microscopy examination. For microleakage evaluation, teeth were coated with nail polish and immersed in dye solution. Teeth were cut in 3 slices and dye penetration was recorded (mm), digitized and analyzed with Image Tool software. Microleakage data were analyzed statistically by non-parametric Kruskal-Wallis and Mann-Whitney tests. RESULTS: Leakage in enamel was lower than in dentin (p<0.001). G2 exhibited the lowest leakage values (p<0.05) in enamel margins, with no differences between the other groups. In dentin margins, groups G1 and G2 had similar behavior and both showed less leakage (p<0.05) than groups G3 and G4. SEM micrographs revealed different marginal adaptation patterns for the different techniques and for the different substrates. CONCLUSION: The soft-start technique showed no leakage in enamel margins and produced similar values to those of the conventional (control) technique for dentin margins.