Cerebrolysin (Cere) is a compound with neurotrophic activity shown to be effective in Alzheimer's disease in earlier trials. The efficacy and safety of three dosages of Cere were investigated in this randomized, double-blind, placebo-controlled, study. Two hundred and seventy-nine patients were enrolled (69 Cere 10 ml; 70 Cere 30 ml; 71 Cere 60 ml and 69 placebo). Patients received iv infusions of 10, 30, 60 ml Cere or placebo 5 days/week for the first 4 weeks and thereafter, two iv infusions per week for 8 weeks. Effects on cognition and clinical global impressions were evaluated 4, 12 and 24 weeks after the beginning of the infusions using the CIBIC+ and the modified Alzheimer's Disease Assessment Scale (ADAS)-cog. At week 24, significant improvement of cognitive performance on the ADAS-cog (P=0.038) and global function (CIBIC+; P>0.001) was observed for the 10 ml dose. The 30 and 60 ml doses showed significant improvement of the global outcome but failed to show significant improvement of cognition. The results are consistent with a reversed U-shaped dose-response relationship for Cere. The percentage of patients reporting adverse events was similar across all study groups. Cere treatment was well tolerated and led to significant, dose-dependent improvement of cognition and global clinical impression.