Lesions in knee joint articular cartilage (AC) have limited repair capacity. Many clinically available treatments induce a fibrous-like cartilage repair instead of hyaline cartilage. To induce hyaline cartilage repair, we hypothesized that type I collagen scaffolds with fibers aligned perpendicular to the AC surface would result in qualitatively better tissue repair due to a guided cellular influx from the subchondral bone. By specific freezing protocols, type I collagen scaffolds with isotropic and anisotropic fiber architectures were produced. Rabbits were operated on bilaterally and two full thickness defects were created in each knee joint. The defects were filled with (1) an isotropic scaffold, (2) an anisotropic scaffold with pores parallel to the cartilage surface, and (3) an anisotropic scaffold with pores perpendicular to the cartilage surface. Empty defects served as controls. After 4 (n=13) and 12 (n=13) weeks, regeneration was scored qualitatively and quantitatively using histological analysis and a modified O'Driscoll score. After 4 weeks, all defects were completely filled with partially differentiated hyaline cartilage tissue. No differences in O'Driscoll scores were measured between empty defects and scaffold types. After 12 weeks, all treatments led to hyaline cartilage repair visualized by increased glycosaminoglycan staining. Total scores were significantly increased for parallel anisotropic and empty defects over time (p