This study was designed to evaluate the potential use of a prototype collagen-chondrocyte allograft in the repair of full-thickness articular cartilage defects. Articular cartilage was harvested from young donor New Zealand White rabbits, enzymatically digested, cultured in monolayer, and passed into a three-dimensional porous type I collagen sponge (American Biomaterials). The composite grafts were incubated for 1 week. (Phase I) Twenty adult NZW rabbits underwent bilateral knee arthrotomies. Three-millimeter full-thickness articular cartilage defects were made in the trochlea of the distal femur. A 4-mm circular punch from the composite cell-seeded grafts was press-fit into the right knee defects. The left knee served as a control (collagen sponge alone or ungrafted defect). Animals were allowed free activity postoperatively and were killed in groups of five at 4, 8, 12, and 24 weeks postoperatively. Defect areas were harvested. Sections were cut at 5-microm thickness and stained with hematoxylin and eosin. The degree as well as quality of healing were assessed and scored with a grading system modified from Salter and O'Driscoll for cartilage repair. (The maximum score was 24 points.) Safranin-O staining as well as polarized light examination of representative sections was undertaken to assess the proteoglycan content and structural characteristics of the repair matrix. (Phase II) An additional 15 NZW rabbits underwent the above procedure but with the addition of fibroblast growth factor (FGF) (100 ng/ml) and insulin (5 microg/ml) to the growth medium of the composite grafts as stimulators of chondrocyte proliferation and proteoglycan synthesis. Control specimens in phase I and II (collagen sponge alone or ungrafted defects) healed with a primarily fibrous or fibrocartilagenous matrix. Defects grafted with cell-seeded collagen sponges demonstrated enhanced healing at all time points examined when compared to controls. There was a strong tendency toward a hyaline appearing matrix with increased Safranin-O staining and birefringence under polarized light more closely resembling the normal native cartilage. Mean histologic score for grafted defects was 18.4 (+/-3.1). Mean scores for collagen sponge alone and ungrafted defects in phase I were 12.7 (+/-4) and 12.7 (+/-3.1) (P<0.01). The addition of FGF and insulin to the growth medium (phase II) resulted in a significantly enhanced repair matrix when compared to the non-FGF-enhanced grafts, with a greater percentage of hyaline appearing tissue at all time points examined (4,8, and 12 weeks). Organization of the chondrocytes was improved at all time points examined as well. Mean histologic score for the FGF-grafted defects was 21.1 (+/-3.0). Mean scores for collagen sponge alone and ungrafted defects in phase II were 14.9 (+/-2.9) and 15.5 (+/-1.9) (p<0.01).