Background and AimRemimazolam tosilate (RT) is under evaluation as a sedative for endoscopic procedures. Herein, we aimed to evaluate safety including cognition recovery of RT administered in elderly patients undergoing upper gastrointestinal endoscopy and assess its safety dosage.MethodsNinety‐nine patients presenting for upper gastrointestinal endoscopy were randomized to receive 0.1 mg/kg RT (R1) or 0.2 mg/kg RT (R2), or propofol (P). Cognitive functions (memory, attention, and executive function) were measured via neuropsychological tests conducted before sedation and 5 min after recovery to full alertness. Adverse events were also assessed.ResultsThere were no statistical differences between postoperative and baseline results for R1 group and P group, whereas those for R2 group revealed worsened postoperative cognitive functions (immediate recall and short delay recall) than baseline (P < 0.05). Compared with P group, Scores demonstrated worse restoration of immediate recall in R1 group, immediate recall, short‐delayed recall, and attention function in R2 group (P < 0.05). Patients in R2 group had a longer sedation time (12.09 vs 8.27 vs 8.21 min; P < 0.001) and recovery time (6.85 vs 3.82 vs 4.33 min; P < 0.001) than that in R1 group and P group. Moreover, the incidence of hypotension was 3.0% in R1 group, whereas it was 21.2% in R2 group and 48.5% in P group (P < 0.05).ConclusionThe addition of 0.1 mg/kg RT as an adjunct to opiate sedation for upper gastrointestinal endoscopy not only achieves more stable perioperative hemodynamics but also achieves acceptable neuropsychiatric functions in elderly patients.