We report the identification of a cell population that shares pericyte, stromal and stemness features, does not harbor the KrasG12D mutation and drives tumoral growth in vitro and in vivo. We term these cells pericyte stem cells (PeSCs) and define them as CD45- EPCAM- CD29+ CD106+ CD24+ CD44+ cells. We perform studies with p48-Cre;KrasG12D (KC), pdx1-Cre;KrasG12D ;Ink4a/Arffl/fl (KIC) and pdx1-Cre;KrasG12D ;p53R172H (KPC) and tumor tissues from PDAC and chronic pancreatitis patients. We also perform single-cell RNAseq analysis and reveal a unique signature of PeSC. Under steady-state conditions, PeSCs are barely detectable in the pancreas but present in the neoplastic microenvironment both in humans and mice. The coinjection of PeSCs and tumor epithelial cells leads to increased tumor growth, differentiation of Ly6G+ myeloid-derived suppressor cells, and a decreased amount of F4/80+ macrophages and CD11c+ dendritic cells. This population induces resistance to anti-PD-1 immunotherapy when coinjected with epithelial tumor cells. Our data reveal the existence of a cell population that instructs immunosuppressive myeloid cell responses to bypass PD-1 targeting and thus suggest potential new approaches for overcoming resistance to immunotherapy in clinical settings.Keywords: PD-1 therapy; myeloid-derived suppressor cells; pancreatic cancer; pericyte stem cells.