Objectives To determine whether the Mini‐Cog can be applied by nursing staff to hospitalized elderly patients for cognitive impairment associated risk stratification. Methods This explorative prospective multicenter cohort study was carried out among 2522 patients aged 70 and older, hospitalized due to physical illness in eight hospitals in Rhineland‐Palatinate, Germany. All patients were asked to conduct the Mini‐Cog at the day of admission and were clustered into low‐performance, intermediate‐performance, and good‐performance categories by trained nursing staff and two experienced geronto‐psychiatrists as gold standard. Complications in the course of the treatment were monitored. Results The Mini‐Cog was conducted in 1398 (54%) out of 2522 eligible patients. Mini‐Cog scores assessed by nursing staff differed from the gold standard in 327 cases (23.9%). According to the area under the curve (AUC), nursing staff identified cognitively low‐performing patients almost as well as the geronto‐psychiatrists (AUC = 0.862; 95% CI, 0.83‐0.89; P < 0.001, accuracy 89.6%). Overall, 241 (17.6%) patients were classified as low performing. These patients had a significantly higher probability of suffering from at least one complication (odds ratio [OR] = 3.13; 95% CI, 2.09‐4.70; calculated by a logistic regression model, adjusted for age), and they had a higher probability to show behavioral symptoms. Conclusion Even under naturalistic conditions, nursing staff detected cognitively low‐performing inpatients with the Mini‐Cog. Using this short screening instrument should enable to predict complications of hospitalized older patients associated with cognitive impairment, a precondition to implement targeted care for this vulnerable patient group.