Objective: Osteosarcoma (OS) is an aggressive, highly metastatic, relatively drug-resistant bone tumor with poor long-term survival rates. The presence and persistence of circulating tumor cells (CTCs) in the peripheral blood are believed to be associated with treatment inefficiency and distant metastases. A blood-based CTC test is thus greatly needed for monitoring disease progression and predicting clinical outcomes. However, traditional methods cannot detect CTCs from tumors of mesenchymal origin such as OS, and research on CTC detection in mesenchymal tumors has been hindered for years. Methods: In this study, we developed a CTC test based on hexokinase 2, a metabolic function-associated marker, for the detection and surveillance of OS CTCs, and subsequently explored its clinical value. Twelve patients with OS were enrolled as the training cohort for serial CTC tests. Dynamic CTC counting, in combination with therapy evaluation and post-treatment follow-up, was used to establish a model for predicting post-chemotherapy evaluation and disease-free survival, and the model was further validated with a cohort of 8 patients with OS. Results: Two dynamic CTC number patterns were identified, and the resulting predictive model exhibited 92% consistency with the clinical outcomes. This model suggested that a single CTC test has similar predictive power to serial CTC analysis. In the validation cohort, the single CTC test exhibited 100% and 87.5% consistency with therapy response and disease-free survival, respectively. Conclusions: Our non-invasive test for detection and surveillance of CTCs enables accurate prediction of therapy efficiency and prognosis, and may be clinically valuable for avoiding inefficient therapy and prolonging survival.