To estimate clinically relevant changes in functional mobility tests and quantitative gait measures at group and individual level in frail elderly patients. This study was a cohort study of consecutively admitted frail elderly patients. Gait velocity, Timed-Up-and-Go test (TUG), and other mobility tests were measured on admission and 2 weeks later. In between, patients received multidisciplinary treatment. Three experts decided from video recordings if patients had a clinically relevant change in gait, defined as change in the expected risk of falling. A total of 85 patients (mean age 75.8 years, 46 female) participated. Of whom, 45% had dementia; 59 patients were stable and 26 showed a clinically relevant change in gait. Gait velocity and TUG were most sensitive to change at group level. In individual patients, a 5% change from baseline in gait velocity and 9% change in TUG had a sensitivity of 92% and 93% for detection of clinically relevant change, but specificity of 27% and 34%, respectively. At group level, gait velocity and TUG were, from all investigated tests, most sensitive to change and in this perspective the best outcome variables. In individual patients, the high intraindividual variability makes these measures unsuitable as independent screening instruments for clinically relevant changes in gait.