Purpose: To evaluate the microtensile bond strength (µTBS) and nanoleakage (NL) of several universal adhesives to eroded dentin (ED), using etch-and-rinse (ER) or self-etch (SE) strategies, and to characterize the surface using two pH cycling models to erode dentin (citric acid and a soft drink). Materials and Methods: Molars were eroded either by soft-drink or citric acid cycling, or were left untreated as control (SD). For each surface, the following adhesives were applied: 1. All-Bond Universal; 2. Ambar Universal; 3. Clearfil Universal; 4. Futurabond U; 5. One Coat 7 Universal; 6. Peak Universal Bond; 7. Prime&Bond Elect; 8. Scotchbond Universal; 9. Tetric n-bond Universal, and 10. Xeno Select. After application of the composite, specimens were sectioned into composite-dentin sticks and tested under tension (0.5 mm/min). Selected sticks from each tooth were used to assess NL. The occlusal dentin surfaces after erosive cycling were examined using SEM. Data were analyzed by three-way ANOVA and Tukey's post-hoc test (a = 0.05). Results: In ED, there was no difference in μTBS and NL between ER and SE strategies (p > 0.61). Most μTBS and NL values obtained for ED were, respectively, lower and higher than those for SD (p 0.01), being worse for citric acid ED (p 0.001). Citric-acid-eroded dentin showed more enlarged tubules, with partial loss of peritubular dentin when compared to soft-drink eroded dentin. Conclusion: The different pH cycling models reduced μTBS and increased NL of the composite/eroded-dentin interface; however, in ED, the performance of the universal adhesives did not depend on the adhesive strategy used. Keywords: tooth erosion, universal adhesive, bond strength, nanoleakage, scanning electron microscopy