Prosthetic materials currently used to repair abdominal wall defects occasionally must be placed in direct contact with the visceral peritoneum. The prosthesis-peritoneum interface is the site of several possible problems, including the formation of adhesions and erosion of the intestinal loops, which may lead to the formation of fistulas. This investigation was designed to compare the behavior of two prosthetic biomaterials in composite form at the level of the peritoneum. Defects (7 x 5 cm) were created in the abdominal wall of 18 white New Zealand rabbits weighing approximately 2500 g. The defects (involving aponeurotic and muscular planes and the parietal peritoneum) were repaired with polypropylene (PL) + ePTFE (Preclude dura substitute) or Parietex composite (PC) prostheses. The prostheses were secured to the edges of the defect by continuous PL sutures interrupted at the corners of the implant. Three study groups were established according to the type of implant: group I (n = 6) (controls)--PL; group II (n = 6)--PL + ePTFE; and group III (n = 6)--PC. The animals were sacrificed 14 days after implant, and the prostheses were examined by light microscopy and scanning electron microscopy (SEM). The formation of adhesions at the prosthesis-visceral peritoneum interface were quantified according to a protocol previously described by us. The biomechanical resistance of the implant was evaluated using strips comprising prosthetic material and anchorage tissue. The Mann-Whitney U-test was used to compare data corresponding to each group. There was no postimplant mortality. No infection or rejection of the prosthesis was observed in any of the animals. Firm adhesions were detected in the PL implants, whereas in the PL + ePTFE and PC implants the adhesions were loose. The mean prosthetic surface areas covered by adhesions were 7.67, 0.10 and 0.19 cm2 for groups I, II, and III, respectively, showing a significant difference between values corresponding to groups I and II and to groups I and III (p < 0.05). Comparison of values recorded for groups II and III yielded no significant difference (p > 0.05). In groups II and III, the neoperitoneum was homogeneous and composed of organized and vascularized connective tissue covered by a mesoendothelium that was interrupted by accumulations of fibroblasts and white blood cells. In contrast, a disorganized neoperitoneum of rough texture was observed in the group I specimens. At times, areas of hemorrhage and necrosis corresponding to the sites of adhesion formation could be observed. Resistance to traction of composite implants (mean +/- SD: 15.72 +/- 1.32 and 15.89 +/- 2.73) was similar to that of the PL implants (15.03 +/- 2.92) (Mann-Whitney U-test, p < 0.05). It may be concluded that (1) composite prostheses show optimum behavior in terms of adhesion formation at the prosthesis-visceral peritoneum interface; (2) the neoperitoneum formed after the implant of a composite prosthesis almost physically and functionally replaces the normal peritoneum; (3) a significantly greater degree of peritoneal regeneration is achieved after implant of a PC prosthesis; and (4) there was no significant difference regarding biomechanical resistance between PL prostheses and PL + ePTFE and Parietex composites.