This study compared the impact of four test methods on adhesion of resin composite to enamel and dentin. Human molars (N = 54) were randomly assigned to test the adhesion of resin composite material (Quadrant Universal LC) using one of the following test methods: (a) macroshear test (SBT; n = 16), (b) macrotensile test (TBT; n = 16), (c) microshear test (mu SBT; n = 16) and (d) microtensile test (mu TBT; n = 6). In a randomized manner, buccal or lingual surfaces of each tooth, were assigned as enamel or dentin substrates. Enamel and dentin surfaces were conditioned using an etch-and-rinse adhesive system (Syntac Classic). After storage (24 h, 37 degrees C), bond tests were conducted in a Universal Testing Machine (1 mm/min) and failure types were analyzed. Data were analyzed using Univariate and Tukey's, Bonneferroni tests (a = 0.05). Two-parameter Weibull modulus, scale (m) and shape (0) were calculated. Test method (p < 0.001) and substrate type (p < 0.001) significantly affected the results. When testing adhesion of resin composite to enamel, SBT (25.9 +/- 5.7)(a), TBT (17.3 +/- 5.1)(a,c) and mu SBT (27.2 +/- 6.6)(a,d) test methods showed significantly higher mean bond values compared to mu TBT (10.1 +/- 4.4)(b) (p < 0.05). Adhesion of resin composite to dentin did not show significant difference depending on the test method (12 +/- 5.7-20.4 +/- 4.8; p > 0.05). Only with SBT, significant difference was observed for bond values between enamel (25.9 +/- 5.7) and dentin (12 +/- 5.7; p < 0.05). Weibull distribution presented the highest shape values for enamel-mu SBT (29.7) and dentin-mu SBT (22.2) among substrate-test combinations. Regardless of the test method, cohesive failures in substrate were more frequent in enamel (19.1%) than in dentin (9.8%).