Purpose: A major concern for implants that penetrate stratified epithelia is aggressive epithelial proliferation and migration. This epithelial downgrowth on the implant can be inhibited by a firm attachment between the underlying connective tissue and the implant. This study evaluates the connective tissue attachment to titanium implants with various well-defined surface topographies. Materials and Methods: Titanium-coated epoxy replicas of polished (PO; Ra = 0.06 µm), finely blasted (FB; Ra = 1.36 µm), coarsely blasted (CB; Ra = 5.09 µm), acid-etched (AE; Ra = 0.59 µm), coarsely blasted and acid-etched (SLA; Ra = 4.39 µm), titanium plasma-sprayed (TPS; Ra = 5.85 µm), machined-like (ML; Ra=2.15 µm), and micromachined grooved (GR; V-shaped grooves 30 µm deep) surfaces were implanted subcutaneously in 74 rats for 1 to 11 weeks. Animals were sacrificed weekly. Surfaces were processed for histomorphometric evaluation of connective tissue attachment, capsule thickness, and where applicable, the degree of separation between the tissue and implant. Results: A total of 153 test surfaces were analyzed. Statistical analysis revealed that textured and rough substrata, namely the GR, TPS, AE, CB, and SLA surfaces, exhibited significantly greater (P .05) connective tissue attachment and thinner fibrous encapsulation when compared to the PO surface. Tissue separation from the implant interface was of significantly lower magnitude and frequency with the rough surfaces than with the PO surface. Conclusions: The results indicate that rough implant surfaces are associated with stable connective tissue attachment, which has implications for their use in percutaneous and permucosal applications. In addition, data from the AE surface may indicate that the geometry of the surface irregularities can also be a significant determinant of the connective tissue response. (Basic Science) (More than 50 references) Palabras clave: connective tissue attachment, fibrous capsule, surface topography, tissue-implant interface