This study evaluated the influence of the dental substrates obtained after the use of different caries removal techniques on bonding of a self-etching system. Forty, extracted, carious, human molars were ground to expose flat surfaces containing caries-infected dentine surrounded by sound dentine. The caries lesions of the specimens were removed or not (control--G1) either by round steel burs and water-cooled, low speed, handpiece (G2), or by irradiation with an erbium, chromium:yttrium scandium gallium garnet (Er,Cr:YSGG) laser (2W, 20 Hz, 35.38 J/cm(2), fiber G4 handpiece with 0.2826 mm(2), non-contact mode at a 2 mm distance, 70% air/20% water--G3) or using a chemo-mechanical method (Carisolv--G4). Caries-infected, caries-affected and sound dentines were submitted to a bonding system followed by construction of a resin-based composite crown. Hour-glass shaped samples were obtained and submitted to a micro-tensile bond test. The bond strength data were compared by analysis of variance (ANOVA), complemented by Tukey's test (P <or= 0.05). The samples of sound dentine presented higher bond strengths than did samples of caries-affected dentine, except for the groups treated with the Er,Cr:YSGG laser. The highest bond strengths were observed with the sound dentine treated with burs and Carisolv. The bond strengths to caries-affected dentine were similar in all groups. Additionally, bonding to caries-affected dentine of the Er,Cr:YSGG laser and Carisolv groups was similar to bonding to caries-infected dentine. Thus, caries-affected dentine is not an adequate substrate for adhesion. Moreover, amongst the caries removal methods tested, the Er,Cr:YSGG laser irradiation was the poorest in providing a substrate for bonding with the tested self-etching system.