OBJECTIVES: To determine the responsiveness of serial Mini-Mental State Examinations (MMSEs) for the diagnosis and monitoring of delirium in elderly hospital patients.DESIGN: Prospective study.SETTING: University teaching hospital.PARTICIPANTS: One hundred sixty-five people admitted to an acute geriatric service.MEASUREMENTS: Subjects were assessed using the MMSE and the Confusion Assessment Method on hospital Days 1 and 6. Changes in scores were compared between patients who remained free of delirium (n=124) and those who by Day 6 had developed delirium (n=14) or had resolution of delirium present on admission (n=22).RESULTS: A number of measures of responsiveness confirmed that serial MMSE scores were responsive to resolution and to development of delirium. A fall of 2 or more points on the MMSE was the best determinant for detecting development of delirium (93% sensitivity, 90% specificity, positive likelihood ratio (LR)=8.9 (95% confidence interval (CI)=5.2-15.1) and negative LR=0.08 95% CI=0.01-0.53)). A rise of 3 or more points was the best determinant for detecting resolution of delirium (77% sensitivity, 75% specificity, positive LR=3.1 (95% CI=2.1-4.5) and negative LR=0.30 (95% CI=0.14-0.66)).CONCLUSION: The MMSE is responsive to short-term changes in cognitive function in elderly patients. Serial MMSE tests should be helpful in monitoring the development and resolution of delirium in this population.