Objective To explore the effective dose and safety of remimazolam and propofol for painless gastroscopy in patients with cirrhosis. Methods A total of 54 cirrhotic patients who underwent gastroscopy were selected. According to the random number table method, they were divided into two groups: a propofol group (group P, n=30) and a remimazolam group (group R, n=24). After intravenous injection of remifentanil at 0.4 μg/kg, group P was administered with propofol at an initial dose of 1.0 mg/kg, while group R was given remimazolam at an initial dose of 0.1 mg/kg. The dose of the next patient was set by the modified Dixon sequential method. The mean arterial pressure (MAP), heart rate, pulse oxygen saturation (SpO2), respiratory rate, and Modified Observer's Assessment of Alertness/Sedation (MOAA/S) scores at entering into the operating room (T0), before gastroscope insertion (T1), after gastroscope insertion (T2), and at the end of the examination (T3) were recorded in both groups. Adverse reactions such as hypotension, hypoxemia, respiratory depression (breathing rate less than 12 breaths per min), bradycardia, injection pain, nausea and vomiting and vertigo were recorded in both groups. The recovery time and the successful rate of sedation were recorded in both groups. The dose‑response relationship between remimazolam and propofol for painless gastroscopy in patients with cirrhosis was analyzed. The 50% effective dose (ED50), 95% effective dose (ED95) and 95% confidence interval (CI) of remimazolam and propofol for painless gastroscopy in patients with cirrhosis were calculated by Probit regression analysis. Results Group P showed decreases in MAP at T1, T2, and T3 (P<0.05), and in breathing rate at T1 and T2 compared with group R (P<0.05), without statistical difference in heart rate and SpO2 at each time point between the two groups (P>0.05). The incidences of injection pain, respiratory depression and vertigo in group P were higher than those in group R (P<0.05). There was no statistical difference in adverse reactions between the two groups (P>0.05). The recovery time in group R was shorter than that in group P (P<0.05). There was no statistical difference in MOAA/S scores at T0 to T3 between the two groups (P>0.05). The successful sedative rate was 100% in both groups. When propofol was used in combination with remifentanil, the ED50 and 95%CI, and ED95 and 95%CI of propofol were 1.586 (1.458, 1.719) mg/kg and 1.880 (1.738, 2.541) mg/kg, respectively, while the ED50 and 95%CI, ED95 and 95%CI of remimazolam were 0.187 (0.166, 0.214) mg/kg and 0.226 (0.204, 0.394) mg/kg for painless gastroscopy in cirrhotic patients, respectively. Conclusions Compared with propofol, remimazolam exhibits sedative effect for painless gastroscopy in cirrhotic patients in a safer and more effective manner, with light cardiopulmonary inhibition, rapid recovery and less adverse reactions.