The study compared the effect of anesthesia with remimazolam and conventional general anesthesia without remimazolam on the incidence of postoperative delirium (POD) in frail elderly patients scheduled for elective non-cardiac surgery. A total of 606 patients were included, of which 301 received remimazolam (R group) and 305 conventional anesthesia (C group). The incidence of POD during the first three postoperative days was significantly lower in the remimazolam group (39.5%) compared with conventional anesthesia (46.9%) (relative risk 0.84; P = 0.038). The remimazolam group also showed less agitation, better sleep quality, and lower pain scores by the third postoperative day (P < 0.01). Multivariate analysis confirmed that the use of remimazolam was an independent protective factor against POD (Adjusted Odds Ratio 0.68; P = 0.041), while older age and longer duration of anesthesia increased the risk of delirium. Thus, the choice of remimazolam represents a promising strategy to reduce postoperative delirium in frail patients.