Myeloid-derived suppressor cells (MDSC) are induced under diverse pathologic conditions, including neoplasia, and suppress innate and adaptive immunity. While the mechanisms by which MDSC mediate immunosuppression are well-characterized, details on how they develop remain less understood. This is complicated further by the fact that MDSC comprise multiple myeloid cell types, namely monocytes and granulocytes, reflecting diverse stages of differentiation and the proportion of these subpopulations vary among different neoplastic models. Thus, it is thought that the type and quantities of inflammatory mediators generated during neoplasia dictate the composition of the resultant MDSC response. Although much interest has been devoted to monocytic MDSC biology, a fundamental gap remains in our understanding of the derivation of granulocytic MDSC. In settings of heightened granulocytic MDSC responses, we hypothesized that inappropriate production of G-CSF is a key initiator of granulocytic MDSC accumulation. We observed abundant amounts of G-CSF in vivo, which correlated with robust granulocytic MDSC responses in multiple tumor models. Using G-CSF loss- and gain-of-function approaches, we demonstrated for the first time that: 1) abrogating G-CSF production significantly diminished granulocytic MDSC accumulation and tumor growth; 2) ectopically over-expressing G-CSF in G-CSF-negative tumors significantly augmented granulocytic MDSC accumulation and tumor growth; and 3) treatment of naïve healthy mice with recombinant G-CSF protein elicited granulocytic-like MDSC remarkably similar to those induced under tumor-bearing conditions. Collectively, we demonstrated that tumor-derived G-CSF enhances tumor growth through granulocytic MDSC-dependent mechanisms. These findings provide us with novel insights into MDSC subset development and potentially new biomarkers or targets for cancer therapy.