38 Background: Age is an important prognostic factor in oncology. Over 20% of men diagnosed with prostate cancer (PC) are ≥ 75 years old. In the growing elderly population, objective methods for predicting outcomes beyond chronologic age are necessary in order minimize the likelihood of withholding curative treatment when warranted. Herein, we describe and analyze age-related differences in clinico-genomic prognostic indices of aggressiveness in localized PC. Methods: Clinical and genomic data for 8,355 patients from the Decipher Genomic Resource Information Database was obtained. Conventional and genomic prognostic indices including Decipher GC scores, PAM50 molecular subtypes (e.g. luminal A/B or basal) NCCN risk groups and Gleason groups (GG) were stratified by age using multivariable logistic regression analyses (MLRA). Results: With increasing decile of age, we observed a higher proportion of high GG and higher Decipher scores. There was a statistically significant increase in the proportion of patients with high Decipher scores with increasing age among GG1 and GG2 (< 55-10.2%, 30.7%, 55-60-15.4%, 25.6%, 60-65-15.9%, 29.7%, 65-70-16.9%, 28.2%, 70-75-17.9%, 30%, and > 75-20.3%, 37.3%, respectively). Furthermore, the prevalence of the PAM50 luminal B subtype (associated with worse prognosis) increased with age among GG1 and GG2 (< 60-22.2%, 40%, 60-65-29.1%, 41.7%, 65-70-28.2%, 39.2%, 70-75-30%, 43.4%, 75-80-33.5%, 44.3%, > 80-34.2%, 52%, respectively). Among higher grade tumors (GG 3-5), no statistically significant differences between the different age groups were observed. MLRA demonstrated that in addition to higher T stage, PSA and GG, each age decile entailed a 20% increased risk for a high Decipher score (OR 1.2, 95% C.I 1.11-1.3, p < 0.001). Conclusions: Older men with lower grade tumors, as opposed to higher grade tumors, harbored worse disease based on genomic risk models. The accepted paradigm of elderly PC patients being treated conservatively based solely on chronologic age, needs to be changed. We provide evidence suggesting the utility of clinical-genomic characterization for better treatment individualization decisions. (GRID; NCT02609269).