Objectives: The aim of this study was to investigate the feasibility and reliability of functional performance tests in people living with dementia and depression. Method: A cross-sectional study was conducted in a geriatric psychiatry hospital. People living with dementia, depression, and healthy older adults were included. Feasibility of the timed up and go test (TUG), the short physical performance battery (SPPB), and the multisurface obstacle test for older adults (MSOT) was assessed based on valid test executions. Test-retest reliability was evaluated by mean difference (MD), coefficient of variation (CV), standardized MD, intraclass correlation coefficient (ICC), and correlation coefficient ( r_s). Results: A total of 53 people in acute geriatric psychiatry and 21 healthy older adults were included. In people living with dementia (n = 23), feasibility was 65.2% (MSOT), 8.7% (TUG), and 8.7% (SPPB). In people living with depression (n = 30), feasibility was 83.3% (MSOT), 80.0% (TUG), and 46.7% (SPPB). Intraclass correlation coefficients and r_s for the MSOT were high (ICC > 0.70) in both groups. Coefficient of variations of the MSOT were between 10.7% and 18.0% (dementia) and 7.1% and 17.0% (depression). Reliability of the TUG and SPPB was not analyzed in people living with dementia, due to low feasibility. In people living with depression, ICCs and r_s were between 0.86 and 0.87 with CVs of 7.2% (TUG) and 0.69 and 0.95 with CVs of 7.8% and 15.1% (SPPB). Conclusion: Feasibility and reliability of established functional performance tests in acute geriatric psychiatry are limited, especially in people living with dementia. New strategies, for example, sensor-based approaches, may allow measurement of functional performance apart from standardized instruction-based test procedures in this clinical population.