Introduction: Frailty is prevalent among older adults with diabetes mellitus. Elevated serum levels of the soluble receptor for advanced glycation-end products (sRAGE) predict mortality in frail older adults. The evidence that sRAGE is also related to higher mortality in older adults with diabetes mellitus is inconsistent. Therefore, this study explored if frailty status influences the relationship between sRAGE and mortality in older adults with this condition. Methods: We analysed data of 391 participants with diabetes mellitus (median age, 76 years) from four European cohorts enrolled in the FRAILOMIC project. Frailty was evaluated at baseline using Fried’s criteria. Serum sRAGE was determined by ELISA. Participants were stratified by frailty status (n = 280 non-frail and 111 frail). Multivariate Cox proportional hazards regression and Kaplan-Meier survival analysis were used to assess the relationship between sRAGE and mortality. Results: During 6 years of follow-up, 98 participants died (46 non-frail and 52 frail). Non-survivors had significantly higher baseline levels of sRAGE than survivors (median [IQR]: 1,392 [962–2,043] pg/mL vs. 1,212 [963–1,514], p = 0.008). High serum sRAGE (>1,617 pg/mL) was associated with increased mortality in the whole diabetes sample after adjustment for relevant confounders (HR 2.06, 95% CI: 1.36–3.11, p < 0.001), and there was an interaction between sRAGE and frailty (p = 0.006). Accordingly, the association between sRAGE and mortality was stronger in the frail group compared to the non-frail group (HR 2.52, 95% CI: 1.30–4.90, p = 0.006 vs. HR 1.71, 95% CI: 0.91–3.23, p = 0.099, respectively). Likewise, Kaplan-Meier curves showed a significant difference in survival rates between frail participants with high sRAGE and those with low sRAGE (p = 0.001), whereas no survival difference was seen in the non-frail group (p = 0.09). Conclusions: Frailty status influences the relationship between sRAGE and mortality in older adults with diabetes mellitus. Determination of sRAGE in this population could be a useful tool for risk stratification.