Glioblastomas are the most lethal primary brain tumor that frequently relapse or progress as focal masses after radiation, suggesting that a fraction of tumor cells are responsible for the tumor regrowth. The identification of a brain tumor cell subpopulation with potent tumorigenic activity supports the cancer stem cell hypothesis in solid tumors. The goal of this study is to determine a methodology for the establishment of primary human glioblastoma cell lines. Our aim is achieved by taking the following approaches: (i) the establishment of primary glioblastoma cell culture; (ii) isolation of neurospheres derived from glioblastoma primary cultures; (iii) selection of CD133 cells from neurospheres, (iv) formation of subspheres in the CD133-positive population, (v) study of the expression level of GFAP, CD133, Nestin, Nanog, CD34, Sox2, CD44, and CD90 markers on tumor subspheres. Hence, we described a successful method for isolation of CD133-positive cell population and establishment of glioblastoma neurospheres from this primary culture, which are more robust than the ones derived straight from the tumor. Pointed out that the neurospheres derived from glioblastoma primary culture showed 29% more cells expressing CD133 then the ones straight tumor-derived, denoting a higher concentration of CD133-positive cells in the neurospheres derived from glioblastoma primary culture. These CD133-positive fractions were able to further generate subspheres. The subspheres derived from glioblastoma primary culture presented a well-defined morphology while the ones derived from the fresh tumor were sparce and less robust. And the negative fraction of CD133 cells was unable to generate subspheres. The tumor subspheres expressed GFAP, CD133, Nestin, Nanog, CD44, and CD90. Also, the present study describes an optimization of neurospheres/subspheres isolation from glioblastoma primary culture by selection of CD133-positive adherent stem cell.