Background: Subjects with osteoarthritis (OA) are at increased risk for cardiovascular (CV) and all-cause mortality. Whether biomarkers improve outcome prediction in these patients remains to be elucidated. We investigated the association between growth differentiation factor 15 (GDF-15), a novel stress-responsive cytokine, and long-term all-cause mortality among OA patients. Methods: Within the Ulm Osteoarthritis Study, GDF-15 has been measured in the serum of 636 subjects, who underwent hip or knee arthroplasty between 1995 and 1996 (median age 65 years). Results: During a median follow-up of 19.7 years, a total of 402 deaths occurred. GDF-15 was inversely associated with walking distance. Compared to the bottom quartile (Q), subjects within the top quartile of GDF-15 demonstrated a 2.69-fold increased risk of dying (hazard ratio (HR) (95% confidence interval (CI)) 2.69 (1.82–3.96) adjusted for age, sex, BMI, smoking status, localization of OA, diabetes, maximum walking distance, total cholesterol, and cystatin C. Further adjustment for NT-proBNP, troponin I, and hs-C-reactive protein did not change the results appreciably (HR (95%CI) 1.56 (1.07–2.28); 1.75 (1.21–2.55); 2.32 (1.55–3.47) for Q2, Q3, and Q4 respectively, p for trend < 0.001). Conclusions: In subjects with OA, GDF-15 represents a potent predictor of decreased survival over >20 years, independently of conventional CV risk factors, renal, cardiac, and inflammatory biomarkers as well as walking disability, previously associated with increased mortality and lower extremity OA.