Cartilage has limited self-regenerative capacity. Tissue engineering can offer promising solutions for reconstruction of missing or damaged cartilage. A major challenge herein is to define an appropriate cell source that is capable of generating a stable and functional matrix. This study evaluated the performance of culture-expanded human chondrocytes from ear (EC), nose (NC) and articular joint (AC), as well as bone-marrow-derived and adipose-tissue-derived mesenchymal stem cells both in vitro and in vivo. All cells (≥ 3 donors per source) were culture-expanded, encapsulated in alginate and cultured for 5 weeks. Subsequently, constructs were implanted subcutaneously for 8 additional weeks. Before and after implantation, glycosaminoglycan (GAG) and collagen content were measured using biochemical assays. Mechanical properties were determined using stress-strain-indentation tests. Hypertrophic differentiation was evaluated with qRT-PCR and subsequent endochondral ossification with histology. ACs had higher chondrogenic potential in vitro than the other cell sources, as assessed by gene expression and GAG content (p < 0.001). However, after implantation, ACs did not further increase their matrix. In contrast, ECs and NCs continued producing matrix in vivo leading to higher GAG content (p < 0.001) and elastic modulus. For NC-constructs, matrix-deposition was associated with the elastic modulus (R² = 0.477, p = 0.039). Although all cells--except ACs--expressed markers for hypertrophic differentiation in vitro, there was no bone formed in vivo. Our work shows that cartilage formation and functionality depends on the cell source used. ACs possess the highest chondrogenic capacity in vitro, while ECs and NCs are most potent in vivo, making them attractive cell sources for cartilage repair.