To prove the efficiency of a specialized geriatric ward (cognitive geriatric unit, CGU) for patients with a fracture of the proximal femur and additional dementia, we conducted a matched-pair analysis comparing 96 patients with fracture of the proximal femur and additional dementia matched for age, sex, surgical treatment and the degree of cognitive impairment by MMSE score. A total of 48 patients were treated in the CGU, offering extended geriatric assessment, special education of staff, and architecture appropriate for patients with cognitive decline. Target criteria were a gain in the Barthel index and Tinetti score, the length of stay, new admissions to nursing home, the frequency of neuroleptic, antidepressant, and antidementive medication, and the number of specified clinical diagnoses for the dementia syndrome. Length of stay was significantly longer in the CGU. The increase of the Tinetti score was significantly higher in the patients in the CGU, regardless of the length of stay (analysis of covariance: treatment (CGU/non-CGU): F(1/93) = 9.421, p = 0.003; covariate (length of stay): F(1/93) = 3.452, p = 0.066, η(2) = 3.6%). In the intervention group, the number of definite diagnoses concerning the dementia syndrome was also higher. Comparison of drug treatment and the percentage of new admission to a nursing home did not differ between groups. Treatment in a specialized, "cognitive geriatric unit" seems to result in better mobility of demented patients with proximal fractures of the femur.