The aim of this study was to compare the microtensile bond strengths (µTBS) and nanoleakage (NL) of four universal adhesives (UAs) applied in two different etching modes, self-etch (se) or etch-and-rinse (er). The occlusal onethird of 48 freshly extracted intact human molars was removed, and the exposed surfaces were treated with one of the four UAs [Single Bond Universal (SBU), Prime Bond NT (PNT), Peak Universal Adhesive (PUA) or All-Bond Universal (ABU)] in se or er mode, with Clearfil SE Bond (CSE) and Adper Single Bond 2 (ASB) as controls (n=8). After composite build-up, specimens were stored in distilled water (37oC/24 h). Composite resin–dentine beams were prepared (0.9 mm2), and the µTBS test was performed. Data were analyzed using one-way ANOVA and Duncan‘s test (α = 0.05). Additional specimens were processed and examined to assay interfacial NL. The failure patterns of the debonded specimens were evaluated using a stereo electron microscope. One additional tooth was prepared for each group for evaluation of NL. The teeth were immersed in a tracer agent (AgNO3 50% w/v) for 24 h. After longitudinal sectioning and polishing, the generated interfaces were examined under scanning electron microscope (SEM). Statistical analysis indicated that the addition of an etching step significantly affected the µTBS of all UAs, except for SBU, compared to se application modes (p<0.05). The highest µTBS value (38.84±2.07 MPa) was obtained in the SBU er group, and the lowest (15.96±1.21 MPa) in the ABU se group. While no NL was determined in the SBU adhesive se or er groups, varying levels were detected in the other groups. The results show that UA performance was material-dependent. Application of an etching step prior to UAs increases their bonding effectiveness and reduces levels of NL.