Treatment of dementing illnesses is still unsatisfactory and agents that might have potential for improving cognitive functions are therefore under extensive investigation. Over a 3-month period a multicenter, double-blind, placebo-controlled, parallel-group study was conducted in 190 patients with dementia to examine whether administration of the xanthine derivative propentofylline at 0.9 g/d might either improve cognition or slow the progression of the disease. Improvement in performance on the Gottfries-Brane-Steen scale (GBS) and in at least two of eight psychometric assessments were defined a priori as the primary efficacy measures. The total scores on the GBS decreased in both the propentofylline and the placebo groups, with improvement however being significantly (p < 0.0001) higher in the propentofylline group. The mean scores in the Mini-Mental State Examination increased significantly (p < 0.05) in both groups, but the increase was significantly (p < 0.02) higher in the propentofylline-treated group. In contrast, psychometric assessments and the cognitive difficulties scale failed to demonstrate a therapeutic benefit of propentofylline after three months. Performance in these categories showed a significant (p < 0.05) improvement for the propentofylline-treated patients and for the placebo-treated patients, but was comparable for both groups. Thus, in spite of the positive results in GBS, the efficacy of propentofylline could not be established according to the definition of the primary efficacy measures. The clinical parameters, however, can be positively influenced by propentofylline. In addition, the results indicate that patient performance can already be enhanced if subjects are required to engage in cognitive tasks.