Objective To assess the feasibility and effectiveness of acute gastrointestinal injury grade (AGI) in elderly patients with sepsis. Methods Elderly patients with sepsis admitted to our ICU between March 2014 and December 2015 and staying for critical care >2 days were enrolled. Age, gender, APACHE Ⅱ score, SOFA score and AGI grade(AGI_initial) at admission, the highest AGI grade(AGI_max) within the first week, length of ICU stay, usage of mechanical ventilation and renal replacement therapy,  28-day mortality and 90-day mortality were recorded. Variables were compared among the AGI grades. Results 91patients were enrolled. There was a significantly difference of SOFA score among AGI_initial grade Ⅰ,Ⅱ and Ⅲ (P＜0.05): the scores were 9.7±3.1,11.3±3.7,13.6±3.7, respectively. SOFA score in AGI_max grade Ⅰ, Ⅱ and Ⅲ+ Ⅳ also shown significantly difference with the score of 8.9±3.4, 10.6±3.5, 12.1±3.6, respectively (P＜0.05). For patients classified as AGI_initial  grade Ⅰ, Ⅱ and Ⅲ, the 28-day mortality rates were 13.2%,46.2% and 78.6%, while the 90-day mortality rates were 23.7%,71.8%,85.7%. For patients classified as AGI_max grade Ⅰ, Ⅱ and Ⅲ+Ⅳ, the 28-day mortality rates were 0.0%, 22.5% and 65.8%, respectively, and the 90-day mortality rates were 15.4%, 45.0% and 76.3%. The higher AGI grade was, the higher mortality of patients could be. Conclusion The current AGI grade system is associated with disease severity in elderly patients with sepsis, and can provide a preliminary assessment regarding the prognosis for patients.