Background <p>The impact of perioperative malnutrition on long-term outcomes in elderly revision arthroplasty remains underexplored. This study aimed to (1) evaluate the association between preoperative GNRI and the risk of re-revision; and (2) assess its prognostic value for all-cause mortality.</p> Methods <p>A retrospective cohort of 771 patients aged ≥ 60 years undergoing hip or knee revision arthroplasty (2008–2024) was analyzed. GNRI was calculated preoperatively and assessed as both continuous and categorical variables. Primary outcomes were all-cause re-revision and all-cause mortality. Logistic regression, Cox models, restricted cubic splines, and ROC analysis were applied.</p> Results <p>Lower GNRI was independently associated with increased risks of re-revision (OR per unit decrease = 1.05; <i>P</i> = 0.005) and mortality (HR for GNRI &lt; 101.21 = 2.14; <i>P</i> = 0.020). RCS showed nonlinear relationships, with inflection points around GNRI = 105. GNRI demonstrated moderate predictive performance for both outcomes (AUC: 0.68 for re-revision, 0.65 for mortality).</p> Conclusions <p>Preoperative malnutrition, as measured by GNRI, predicts both re-revision and all-cause mortality in elderly revision arthroplasty patients. Nutritional risk stratification may help guide perioperative management.</p> Level of evidence <p>Level III, prognostic study.</p>