Purpose: To evaluate the 18-month clinical performance of a multimode adhesive (Scotchbond Universal Adhesive, SU, 3M ESPE, St Paul, MN, USA) in noncarious cervical lesions (NCCLs) using two evaluation criteria. Materials and Methods: Thirty-nine patients participated in this study. Two-hundred restorations were assigned to four groups: ERm, etch-and-rinse + moist dentin; ERd, etch-andrinse + dry dentin; Set, selective enamel etching; and SE, self-etch. The composite resin, Filtek Supreme Ultra (3M ESPE), was placed incrementally. The restorations were evaluated at baseline, and at 18 months, using both the World Dental Federation (FDI) and the United States Public Health Service (USPHS) criteria. Statistical analyses were performed using Friedman repeated-measures analysis of variance by rank and McNemar test for significance in each pair (α=0.05). Results: Five restorations (SE: 3; Set: 1; and ERm: 1) were lost after 18 months (p>0.05 for either criteria). Marginal staining occurred in four and 10% of the restorations evaluated (p>0.05), respectively, for USPHS and FDI criteria. Nine restorations were scored as bravo for marginal adaptation using the USPHS criteria and 38%, 40%, 36%, and 44% for groups ERm, ERd, Set, and SE, respectively, when the FDI criteria were applied (p>0.05). However, when semiquantitative scores (or SQUACE) for marginal adaptation were used, SE resulted in a significantly greater number of restorations, with more than 30% of the total length of the interface showing marginal discrepancy (28%) in comparison with the other groups (8%, 6%, and 8%, respectively, for ERm, ERd, and Set). Conclusions: The clinical retention of the multimode adhesive at 18 months does not depend on the bonding strategy. The only differences between strategies were found for the parameter marginal adaptation, for which the FDI criteria were more sensitive than the USPHS criteria.