Objective: To study the enamel- and dentin-resin interfaces of “all-in-one” self-etch adhesives. Methods: 45 extracted human molars were used. Teeth were randomly assigned to one of three substrates: dentin, intact enamel, and roughened enamel. For each substrate specimens were randomly assigned to one of 5 all-in-one adhesives: (1) Adper Prompt L-Pop (AP, 3M ESPE); (2) Clearfil S³ Bond (S3, Kuraray); (3) G-Bond (GB, GC America) (4) iBond (iB, Heraeus Kulzer); (5). Xeno IV (XE, Dentsply). For dentin, a 1 mm thick increment of Filtek Flow (3M ESPE) was applied and light-cured after the application of the respective adhesive on middle dentin. The dentin specimens were fixed, dehydrated, and processed for FESEM and TEM observation. For enamel, after the application of the adhesive on the proximal surfaces, specimens were immersed in acetone and left in a rotator for 4 h. Enamel specimens were then processed for FESEM. Results: For dentin, all the adhesives resulted in a very thin hybridization. iB and GB resulted in a 0.2-0.3µm thick hybrid layer, while that of S3 and XE was close to 0.5µm. AP resulted in the thickest hybrid layer (0.8-1.2µm). For intact enamel GB and iB resulted in featureless morphology resembling that of untreated enamel. XE resulted in areas of intra-prismatic etching and areas without etching characteristics. S3 resulted in frequent shallow intra-prismatic etching, while AP was able to unveil the enamel crystallites across the entire enamel surface. For roughened enamel, iB and GB displayed areas with signs of a mild etching effect intermingled with zones unaffected by the adhesive. XE and S3 unveiled the enamel crystallites in more than 80% of the surface area. AP displayed the classical phosphoric acid etching pattern on the entire surface. Conclusions: There is a wide variation among ‘all-in-one' adhesives in their ability to infiltrate enamel and dentin.