616 Background: Proteomic analysis continues to provide major insight into the pathophysiology of colorectal cancer (CRC). Recently, the Cancer Genome Atlas Project and others have utilized reverse-phase protein arrays (RPPAs) to identify critical protein markers and signaling pathways in a variety of tumor types. However, the prognostic relevance of many of these proteins remains unclear. We utilized RPPA to analyze stage 2 and 3 CRC to investigate the prognostic implications of the functional proteome. Methods: Protein extraction was performed on 232 snap frozen stage 2/3 samples from the MD Anderson Cancer Center with a median 5 year follow up. 163 validated proteins and phospho-proteins were analyzed by RPPA (with dichotomization by the median value), and used to identify protein predictors of tumor recurrence. Cox regression was used for univariate analysis with bootstrap validation, followed by inclusion of proteins with corrected p≤ 0.05 into multivariate model with clinical and pathologic variables, including stage, grade, and microsatellite status. Results: 12 proteins were found to be significant predictors of tumor recurrence on univariate analysis after bootstrap validation, which are notable for components of the energy balance/MTOR signaling pathway including AMPK, mTOR, PI3 Kinase p85, FoxO3a. On multivariate analysis, inclusive of known prognostic clinical and pathology variables, phospho-Bad (Ser112), FoxO3a, HER3, and phospho-S6 (Ser240-244) remained significant. Conclusions: Functional proteomic analysis has identified key proteomic features with prognostic importance independent of known clinical/pathological variables. Confirmation studies are ongoing to explore regulators of energy balance and apoptosis in colorectal cancer. [Table: see text]