The primary substratum for epithelial migration during marsupialization of percutaneous implants is adjacent connective tissue. The purpose of this investigation was to test the hypothesis that if the latter is composed of deep connective tissue or other deep tissue, it will inhibit this process. Onehundred-forty-two smooth-surfaced polyethylene implants of a simple geometric design were implanted in dorsal skin of 6-8-week-old male C57/BL/6 mice such that the stems passed through and deep to the panniculus carnosus (PC), which is a deep tissue. Animals were sacrificed in groups between 1 and 11 weeks, and the implants and surrounding tissue were harvested, embedded in Historesin, and 3-pm thick step serial sections prepared. No implants were lost by extrusion and 133 were suitable for examination. Epithelial investment of the devices proceeded deep to PC around only five of the implants over the time course. In the remainder, the mean distance of the advancing epithelial edge from the upper border of the PC was 0.25 mm by 1 week and this did not change significantly thereafter ( p = 0.647). Significant extrusion of all devices occurred ( p = 0.0001). It was concluded that these results are consistent with the hypothesis that deep connective tissue, such as the fascia surrounding the PC or muscle tissue such as the PC itself, can act as a functional barrier to marsupialization, and that exfoliation of such devices is not an epithelial function.