The uncompetitive N-methyl-D-aspartate (NMDA) antagonist memantine was tested against placebo in two randomized controlled trials. In total, 900 patients suffering from mild-to-moderate “probable” VaD (according to NINDS-AIREN criteria) were included. In these prospective, 2-arm parallel, multicenter trials conducted in the United Kingdom (MMM500) and in France (MMM300), patients suffering from “probable” vascular dementia (according to NINDS-AIREN criteria) were recruited. Active treatment in both trials was memantine at the standard daily dose of 10 mg b.i.d. The cognitive subscale of the Alzheimer's Disease Assessment Scale (ADAS-cog) was a primary endpoint in both trials, and in both trials a statistically significant difference was seen between treatment groups after 28 weeks. In a pooled analysis of the data, various subgroups were examined. In a first analysis, patients were stratified by their severity of dementia (measured by the MMSE total scores at baseline). In this analysis, memantine was superior to placebo in all subgroups, but the magnitude of effect was clearly more pronounced in the more severely demented patients. A second analysis stratified the patients by the neuroradiological findings at baseline (“small vessel” versus “large vessel” type of VaD). The cognitive benefit by memantine treatment was larger for the small vessel group and, interestingly, also the decline in the placebo group was faster in the small vessel patients. In these trials, memantine at a dose of 10 mg b.i.d. was safe and very well tolerated with a frequency of dropouts due to adverse events that was close to placebo.