Survival rates of osteosarcoma patients could not be significantly improved by conventional chemotherapeutic treatment regimens since the introduction of high-dose chemotherapy 35 years ago. Therefore, there is a strong clinical need for new therapeutic targets and personalized treatment strategies, requiring reliable in vivo model systems for the identification and testing of potential new treatment approaches. Conventional in vivo rodent experiments face ethical issues, are time consuming and costly, being of particular relevance in orphan diseases like osteosarcoma. An attractive alternative to such animal experiments is the chicken chorioallantoic membrane (CAM) assay. The CAM is a highly vascularized, non-innervated extra-embryonic membrane that is perfectly suited for the engraftment of tumor cells. However, only few reports are available for osteosarcoma and reported data are inconsistent. Therefore, the aim of this study was the adaptation and optimization of the CAM assay for its application in osteosarcoma research. Tumor take rates and volumes of osteosarcoma that developed on the CAM were analyzed after modification of several experimental parameters, including egg windowing, CAM pretreatment, inoculation technique and many more. Eight osteosarcoma cell lines were investigated. Our optimized OS-CAM-assay was finally validated against a rat animal xenograft model. Using the cell line MNNG HOS as reference we could improve the tumor take rates from 51% to 94%, the viability of the embryos from initially 40% to >80% and achieved a threefold increase of the tumor volumes. We were able to generate solid tumors from all eight osteosarcoma cell lines used in this study and could reproduce results that were obtained using an osteosarcoma rat animal model. The CAM assay can bridge the gap between in vitro cell culture and in vivo animal experiments. As reliable in vivo model for osteosarcoma research the optimized CAM assay may speed up preclinical data collection and simplifies research on potential new agents towards personalized treatment strategies. Further, in accordance with Russell's and Burch's "Principles of Humane Experimental Technique" the reasonable use of this model provides a refinement by minimizing pain and suffering of animals and supports a considerable reduction and/or replacement of animal experiments.