Autologous chondrocyte implantation is currently applied in clinics as an innovative tool for articular cartilage repair. Animal models have been and still are being used to validate and further improve the technique. However, in various species, the outcome varies from hyaline-like cartilage to fibrocartilage. This may be due partly to the spontaneous dedifferentiation of chondrocytes once cultured in vitro. Here we assessed whether the extent of dedifferentiation varies between species and we hypothesized that the level of chondrocyte phenotype stability during expansion may contribute to the maintenance of their chondrogenic commitment and redifferentiation potential. Condyle chondrocytes were harvested from sheep, dog, and human, and expanded for 1, 6, or 12 cell duplications. At each interval, cell phenotype was monitored (morphology and biosynthesis of cartilage markers) and redifferentiation was assessed by an in vitro assay of chondrogenesis in micromass pellet and an in vivo assay of ectopic cartilage formation in immunodeficient mice. Results indicate that, during culture, the sheep chondrocyte phenotype is maintained better than that of human chondrocytes, which in turn dedifferentiate to a lesser extent than dog chondrocytes Accordingly, after expansion, sheep chondrocytes spontaneously reform hyaline-like cartilage; human chondrocytes redifferentiate only under stimulation with chondrogenic inducers whereas, after a few passages, dog chondrocytes lose any capacity to redifferentiate regardless of the presence of inducers. Thus, conditions allowing cartilage formation in one species are not necessarily transposable to other species. Therefore, results with animal models should be cautiously applied to humans. In addition, for tissue-engineering purposes, the number of cell duplications must be, for each species, carefully monitored to remain in the range of amplification allowing redifferentiation and chondrogenesis.