Introduction: With an aging population, understanding how frailty affects treatment response is paramount to improve patient-centered outcomes. Hypothesis: We hypothesized that circulating proteomic correlates of frailty will identify novel pathways for therapeutic investigation and provide prognostic information. Methods: From a multicenter, prospective cohort of 809 participants (median age 83 years; 44% women) with symptomatic, severe aortic stenosis (AS) and systematic assessments of frailty, we measured >900 circulating proteins (Olink). We used principal component analysis (PCA) to define composite “axes” of 12 measures of frailty (Figure). Using ordinary and penalized regression (LASSO), we related proteins to both individual measures of frailty and the composite axes. Prioritized proteins were then examined for their relation with age and morbidity in external, published datasets. LASSO model coefficients were used to calculate protein “scores” of frailty, which were tested for their relation with all-cause mortality in the AS cohort and cause specific mortality in >1,800 participants from the Framingham Heart Study (FHS; mean age 55 years, 54% women). Results: PCA identified 3 axes of frailty (“patient reported outcomes”, “body composition” and “physical function”) that explained ≈49% of variance. Proteins related to axes of frailty were (1) related to inflammatory and metabolic phenotypes and (2) weakly related to age, in >35,000 community dwelling Icelanders (mean age 55±17 years). The protein score of physical function was associated with all-cause post-transcatheter aortic valve implantation (TAVI) mortality in the AS cohort (HR 0.68 [0.59-0.79]) and non-cardiovascular mortality in FHS (0.83 [0.75-0.91]). Conclusions: Circulating proteins capture frailty phenotypes that are weakly related to age, and provide prognostic information in both older adults undergoing TAVI and younger community dwelling adults.