Slow-cycling and treatment-resistant cancer cells escape therapy, providing a rationale for regrowth and recurrence in patients. Much interest has focused on identifying the properties of slow-cycling tumor cells in glioblastoma (GBM), the most common and lethal primary brain tumor. Despite aggressive ionizing radiation (IR) and treatment with the alkylating agent temozolomide (TMZ), GBM patients invariably relapse and ultimately succumb to the disease. In patient biopsies, we demonstrated that GBM cells expressing the proliferation markers Ki67 and MCM2 displayed a larger cell volume compared to rare slow-cycling tumor cells. In optimized density gradients, we isolated a minor fraction of slow-cycling GBM cells in patient biopsies and tumorsphere cultures. Transcriptional profiling, self-renewal, and tumorigenicity assays reflected the slow-cycling state of high-density GBM cells (HDGCs) compared to the tumor bulk of low-density GBM cells (LDGCs). Slow-cycling HDGCs enriched for stem cell antigens proliferated a few days after isolation to generate LDGCs. Both in vitro and in vivo, we demonstrated that HDGCs show increased treatment-resistance to IR and TMZ treatment compared to LDGCs. In conclusion, density gradients represent a non-marker based approach to isolate slow-cycling and treatment-resistant GBM cells across GBM subgroups.