Forty geriatric in-patients with severe cognition disorders were randomly allocated to treatment with either 600 mgfipexide daily or placebo over a period of 3 weeks. Before and after treatment, the symptoms of cognitive performance (disorders of memory and attention, asthenia, apathy and disorders of coenaesthesia) were monitored and scored. Similarly, the Thurstone test (symbol matching test) was performed and time to completion, number of errors and exactitude index were recorded. Haemodynamics, haematology and haematochemistry investigations were made before and after treatment, and accessory symptoms of potential side-reactions were monitored by positive questioning. Treatment with fipexide was associated with a significant improvement in each and all monitored symptoms and signs to an average extent of 60%, whereas placebo was not. Similarly, the patients given fipexide experienced a significant improvement in the Thurstone test, in terms of time to completion (-22%), number of errors (-46%) and exactitude index (%60%); again, placebo was not associated with any significant improvement (variations, respectively, of -5%,-14%, and %24%). Overall, 85% of the patients given fipexide experienced clinical improvement to a greater or lesser degree, a significantly greater proportion than that associated with placebo (25%; p<R0.001). Tolerance, both subjective and objective, was good in both treatment groups.