The goal of this study was to correlate prostatic metabolite concentrations from snap-frozen patient biopsies of recurrent cancer after failed radiation therapy with histopathological findings, including Ki-67 immunohistochemistry and pathologic grade, in order to identify quantitative metabolic biomarkers that predict for residual aggressive versus indolent cancer. A total of 124 snap-frozen transrectal ultrasound (TRUS)-guided biopsies were acquired from 47 men with untreated prostate cancer and from 39 men with a rising prostate-specific antigen and recurrent prostate cancer following radiation therapy. Biopsy tissues with Ki-67 labeling index ≤ 5% were classified as indolent cancer, while biopsy tissues with Ki-67 labeling index > 5% were classified as aggressive cancer. The majority (15 out of 17) of cancers classified as aggressive had a primary Gleason 4 pattern (Gleason score ≥ 4 + 3). The concentrations of choline-containing phospholipid metabolites (PC, GPC, and free Cho) and lactate were significantly elevated in recurrent cancer relative to surrounding benign tissues. There was also a significant increase in [PC] and reduction in [GPC] between untreated and irradiated prostate cancer biopsies. The concentration of the choline-containing phospholipid metabolites was significantly higher in recurrent aggressive (≈ twofold) than in recurrent indolent cancer biopsies, and the receiver operating characteristic (ROC) curve analysis of total choline to creatine ratio (tCho/Cr) demonstrated an accuracy of 95% (confidence interval = 0.88-1.00) for predicting aggressive recurrent disease. The tCho/Cr was significantly higher for identifying recurrent aggressive versus indolent cancer (tCho/Cr = 2.4 ± 0.4 versus 1.5 ± 0.2), suggesting that use of a higher threshold tCho/Cr ratio in future in vivo (1)H MRSI studies could improve the selection and therapeutic planning for patients who would benefit most from salvage focal therapy after failed radiation therapy.