BackgroundRecently, albumin–globulin ratio (AGR), a serological indicator that reflects nutritional status and systemic inflammatory, has been reported to be associated with the prognosis of various cancers. However, there is currently no research report on its relationship with cancer cachexia.ObjectivesThis study aimed to explore the prognostic value of AGR in patients with cancer cachexia through a multicenter retrospective analysis.MethodsWe recruited 2,364 patients with cancer cachexia and randomly divided the patients into training and validation cohorts at a ratio of 7:3. The optimal stratification method was used to determine the optimal cutoff value of AGR. The survival curve was evaluated by the Kaplan–Meier method. Cox regression proportional-hazards model was used to determine independent prognostic factors in patients with cancer cachexia. The time-dependent receiver operating characteristic curve was used to compare the prognostic performance of different malnutrition evaluation tools.ResultsThe optimal cutoff value of AGR is 1.24 in patients with cancer cachexia. Increasing AGR was associated with survival in a dose–response manner with a forward L-shape. Compared with the high AGR group, the low AGR group had a shorter overall survival; and there was consistency in training and validation cohorts. In the stratified analysis of TNM stage, AGR has good prognostic distinguishing ability for advanced patients. Multivariate survival analysis determined that low AGR was an independent risk factor affecting all-cause mortality in patients with cancer cachexia. In addition, compared with other malnutrition evaluation tools, AGR could effectively stratify the prognosis of patients with cancer cachexia.ConclusionAGR was an independent prognostic factor affecting patients with cancer cachexia, especially in advanced patients. Compared with other malnutrition evaluation tools, AGR can effectively stratify the prognosis of patients with cancer cachexia.