The purpose of this study was to evaluate the bond strength of three adhesive systems to superficial and deep dentine using the microtensile bond strength test (microTBS). The occlusal enamel of thirty human third molars was removed to expose a flat surface of superficial or deep dentin. For each type of surface, the test specimens were randomly divided into three groups which underwent the application of a conventional two-step adhesive system [Single Bond (SB)] as the control group (n=10), a two-bottle self-etching system [One Coat SE Bond (OCSE)] (n=10) and a one bottle one-step system [Clearfil S3 Bond (CFS3)] (n=10). Adhesives were applied, a 5-mm high "crown" as built-up with resin composite Z250 (3M) and the specimens with a cross-sectional area of 0.7 +/- 0.1 mm2 were tested in tension (0.5 mm/min). Four fractured sticks from each tooth were randomly selected and the dentin side was gently abraded with a 1200-grit SiC paper etched with 35% phosphoric acid for 15 s and air dried. SEM micrographs at 70X and 2400X magnification were taken using scanning electron microscopy (SEM) to calculate the area of tubular dentin (ATD) and tubular density (TD) with Image Pro Plus 5. Two-way ANOVA (dentin depth-adhesive) showed higher bond strength values for SB. However the values did not depend on dentin depth. Linear regression showed a significant relationship between bond strength and area of intertubular dentin for SB (p = 0.004), and a significant inverse relationship between tubular density and bond strength for CFS3 (p = 0.009). OCSE exhibited a tendency that was similar to SB and opposite to CFS3, but was not statistically significant. The conventional two-step adhesive had higher bond strength values. The use of digital image analysis facilitates the manipulation of data and contributes to the interpretation of the behavior of new adhesive systems.