Statement of problem. There are no data available on whether or to what extent hydrofluoric acid affects the marginal integrity of dentin-bonded composite restorations when it is used instead of phosphoric acid in the total-etch technique. Purpose. This in vitro study examined the etching effects of phosphoric acid versus a combination of phosphoric and hydrofluoric acid by evaluation of microleakage in a composite restoration bonded with different dentin adhesive systems. Material and methods. Extracted teeth (n = 90) containing 2 class II preparations, mesial occlusal (MO) and distal occlusal (DO) standarized (cervical margins in dentin) were perfused with Ringer solution and etched in 1 of 2 ways: with phosphoric acid only or with phosphoric combined with hydrofluoric acid. Different dentin bonding agents were then applied (Etch & Prime 3.0, Optibond Solo, Prime & Bond NT, Scotchbond 1, Syntac Single Component, or Syntac Sprint; (n = 15 for each etching material)). The preparations were restored with a hybrid composite (Herculite XRV) and submitted to 5000 thermocycles (5°C to 55°C) to simulate the in vivo situation. Microleakage was assessed with 2% methylene blue diffusion for 24 hours. Dye penetration was calculated as a percentage of the total length of the gingival margins of the preparation with light microscopy at original magnification × 32. The results were analyzed with the Kruskal-Wallis multiple comparison z-value assay (α =.05). Results. Differences in dye penetration were significant, both as a function of the dentin adhesive and the conditioning mode applied. In the specimen groups conditioned with phosphoric acid, Optibond Solo (54% ± 44%) and Syntac Sprint (74% ± 39%) demonstrated the lowest penetration values. Higher values were obtained for Prime & Bond NT (81% ± 34%), Scotchbond 1 (83% ± 31%), Etch & Prime 3.0 (85% ± 33%), and Syntac Single Component (95% ± 16%), with no significant differences (α=.05) between specimen groups. The best results were obtained for Syntac Sprint (24% ± 26% dye penetration) after conditioning with a mixture of phosphoric and hydrofluoric acid. The least favorable result was obtained for Optibond Solo (65% ± 31%). It was significantly different from Prime & Bond NT (76% ± 37%), Scotchbond 1 (85% ± 29%), and Etch & Prime 3.0 (88% ± 24%). Syntac Single Component (75% ± 32%) was significantly different from Syntac Sprint. Syntac Single Component and Syntac Sprint exhibited significantly better results when conditioned with a combination of phosphoric acid and hydrofluoric acid than with phosphoric acid only. Conclusion. Within the limitations of this in vitro study, total-etching water-based (Syntac Single Component) and acetone-based (Syntac Sprint) bonding agents with a combination of phosphoric acid and hydrofluoric acid led to significant reductions (α=.05) in dye penetration compared to phosphoric acid conditioning only. Ethanol-based dentin bonding agents (Etch & Prime 3.0, Optibond Solo, and Scotchbond 1) were not significantly influenced by the type of conditioner used. (J Prosthet Dent 2003;89:161-9.)