High-throughput screening of a natural compound library was performed to identify the most efficacious combinatorial treatment on prostate cancer. Ursolic acid, curcumin and resveratrol were selected for further analyses and administered in vivo via the diet, either alone or in combination, in a mouse allograft model of prostate cancer. All possible combinations of these natural compounds produced synergistic effects on tumor size and weight, as predicted in the screens. A subsequent untargeted metabolomics and metabolic flux analysis using isotopically labeled glutamine indicated that the compound combinations modulated glutamine metabolism. In addition, ASCT2 levels and STAT3, mTORC1 and AMPK activity were modulated to a greater extent by the combinations compared to the individual compounds. Overall, this approach can be useful for identifying synergistic combinations of natural compounds for chemopreventive and therapeutic interventions.