Abstract: Objective : To investigate whether established human glioma cell lines/strains contain tumor stem cells. Methods : Human glioma cell line SHG-44, as well as it's subline SHG-44-9, maintained for long term were cultured in DMEM supplemented with 10% FCS, and in serum-free DMEM/F12 containing bFGF, LIF and EGF respectively. Neural stem cell marker CD133 was used to isolated tumor stem cells by magnetic cell sorting. Brain tumor stem cells were detected before and after the differentiation with flow cytometry and the immunohisto chemistrical staining with antibodies against Hoechst33342, Nestin, NSE and GFAP. Results : Using magnetic cell sorting, CD133 positive cells accounted for 0.021% of SHG-44 cells and 0.035% of SHG-44-9 cells when cultured in DMEM supplemented with 10% FCS, while in serum-free condition, percentage of cells expressing CD133 increased to 1.2% and 2.3% respectively. Flow cytometry revealed that 1.5% of SHG-44 cells and 0.37% of SHG-44-9 cells, cultured in medium with serum, were positive for CDl33; when serum was removed,CD133 positive cell fraction jumped to 16.4% for SHG-44 cells, and 29.1% for SHG-44-9 cells respectively.Moreover, CD133 positive cells can be proliferated and differentiated into neuronal and glial cells. When Nestin, another surface marker, was used, 51.05% SHG-44 cells and 77.53% of SHG-44-9 cells were positive. Conclusions : Human glioma cell lines cultured for long period still contain tumor stem cells, which could be effectivedly isolated both by magnetic cell sorting and by flow cytometry.Nestm positive cells might represent neural precursors or progenitors, so it cannot be used as a specinc marker for brain tumor stem cells.