Levels of vascular endothelial growth factor (VEGF) are regulated, in part, through activation of the phosphatidylinositol 3′-kinase/Akt pathway. Using pharmacologic inhibitors, we have examined the relative contributions of Akt and mammalian target of rapamycin (mTOR) signaling to VEGF production in neuroblastoma and rhabdomyosarcoma cells growing under normoxic (21% O2) or hypoxic (1% O2) conditions. Exogenous VEGF stimulated both Akt and extracellular signal–regulated kinase 1/2 phosphorylation in six of seven rhabdomyosarcoma cell lines but in only one of seven neuroblastoma cells, suggesting autocrine stimulation predominantly in rhabdomyosarcoma cell lines. In general, under normoxic conditions, neuroblastoma cells produced more VEGF (120–1,180 pg/106 cells/24 h) compared with rhabdomyosarcoma lines (0–200 pg/106 cells/24 h). Rapamycin, a selective inhibitor of mTOR, reduced VEGF production in rhabdomyosarcoma cells under normoxic conditions and partially suppressed hypoxia-driven increases in VEGF. However, it poorly inhibited VEGF production under either condition in the majority of neuroblastoma cell lines despite inhibition of mTOR signaling. Rapamycin failed to modulate levels of hypoxia-inducible factor 1α (HIF-1α) under normoxic conditions and modestly reduced hypoxia-driven increases in HIF-1α only in rhabdomyosarcoma cells. In contrast to rapamycin, inhibition of Akt by A-443654 completely blocked signaling to glycogen synthase kinase 3β and had more dramatic effects on VEGF production. Notably, A-443654 significantly inhibited VEGF production in rapamycin-refractory neuroblastoma cell lines. Importantly, whereas combining A-443654 with rapamycin had variable effect on cell proliferation, the combination essentially blocked hypoxia-driven increases in VEGF in all cell lines examined, suggesting that dual blockade at different levels in the phosphatidylinositol 3′-kinase–initiated signaling pathway may be a reasonable strategy for preventing VEGF production in cancer cells derived from pediatric solid tumors. However, this will require formal testing in vivo using animal models of childhood cancer. [Mol Cancer Ther 2007;6(5):1620–8]