Tensile bond strength measurements are commonly used for the evaluation of dentin adhesive systems. Most tests are performed using extracted non-carious human or bovine dentin. However, the adhesion of resins to caries-affected dentin is still unclear. The objectives of this study were to test the hypothesis that bonding to caries-affected dentin is inferior to bonding to normal dentin, and that the quality of the hybrid layer plays a major role in creating good adhesion. We used a micro-tensile bond strength test to compare test bond strengths made to either caries-affected dentin or normal dentin, using three commercial adhesive systems (All Bond 2, Scotchbond Multi-Purpose, and Clearfil Liner Bond II). For scanning electron microscopy, the polished interfaces between the adhesive bond and dentin were subjected to brief exposure to 10% phosphoric acid solution and 5% sodium hypochlorite, so that the quality of the hybrid layers could be observed. Bonding to normal dentin with either All Bond 2 (26.9 +/- 8.8 MPa) or Clearfil Liner Bond II (29.5 +/- 10.9 MPa) showed tensile bond strengths higher than those to caries-affected dentin (13.0 +/- 3.6 MPa and 14.0 +/- 4.3 MPa, respectively). The tensile bond strengths obtained with Scotchbond Multi-Purpose were similar in normal and caries-affected dentin (20.3 +/- 5.5 MPa and 18.5 +/- 4.0 MPa, respectively). The hybrid layers created by All Bond 2 in normal dentin and by Clearfil Liner Bond II in normal or caries-affected dentin showed phosphoric acid and sodium hypochlorite resistance, whereas the hybrid layers created by All Bond 2 in caries-affected dentin and those created by Scotchbond Multi-Purpose to normal and caries-affected dentin showed partial susceptibility to the acid and sodium hypochlorite treatment. The results indicate that the strength of adhesion to dentin depends upon both the adhesive system used and the type of dentin. Moreover, the quality of the hybrid layer may not always contribute significantly to tensile bond strength.