•IP-10 correlates with disease severity and the presence of JAK2 and TET2 mutations•Mouse models have similarly dysregulated IP-10 expression as patients•IP-10 loss partially ameliorates the polycythemia phenotype in mouse models Key studies in pre-leukemic disorders have linked increases in pro-inflammatory cytokines with accelerated phases of the disease, but the precise role of the cellular microenvironment in disease initiation and evolution remains poorly understood. In myeloproliferative neoplasms (MPNs), higher levels of specific cytokines have been previously correlated with increased disease severity (tumor necrosis factor-alpha TNF-α, interferon gamma–induced protein-10 IP-10 or CXCL10) and decreased survival (interleukin 8 IL-8). Whereas TNF-α and IL-8 have been studied by numerous groups, there is a relative paucity of studies on IP-10 (CXCL10). Here we explore the relationship of IP-10 levels with detailed genomic and clinical data and undertake a complementary cytokine screen alongside functional assays in a wide range of MPN mouse models. Similar to patients, levels of IP-10 were increased in mice with more severe disease phenotypes (e.g., JAK2V617F/V617F TET2−/− double-mutant mice) compared with those with less severe phenotypes (e.g., CALRdel52 or JAK2+/V617F mice) and wild-type (WT) littermate controls. Although exposure to IP-10 did not directly alter proliferation or survival in single hematopoietic stem cells (HSCs) in vitro, IP-10−/− mice transplanted with disease-initiating HSCs developed an MPN phenotype more slowly, suggesting that the effect of IP-10 loss was noncell-autonomous. To explore the broader effects of IP-10 loss, we crossed IP-10−/− mice into a series of MPN mouse models and showed that its loss reduces the erythrocytosis observed in mice with the most severe phenotype. Together, these data point to a potential role for blocking IP-10 activity in the management of MPNs.