BACKGROUND Prevalence of Alzheimer's disease in people with Down's syndrome is very high, and many such individuals who are older than 40 years have pathological changes characteristic of Alzheimer's disease. Evidence to support treatment with Alzheimer's drugs is inadequate, although memantine is beneficial in transgenic mice. We aimed to assess safety and efficacy of memantine on cognition and function in individuals with Down's syndrome. METHODS In our prospective randomised double-blind trial, we enrolled adults (>40 years) with karyotypic or clinically diagnosed Down's syndrome, with and without dementia, at four learning disability centres in the UK and Norway. We randomly allocated participants (1:1) to receive memantine or placebo for 52 weeks by use of a computer-generated sequence and a minimisation algorithm to ensure balanced allocation for five prognostic factors (sex, dementia, age group, total Down's syndrome attention, memory, and executive function scales [DAMES] score, and centre). The primary outcome was change in cognition and function, measured with DAMES scores and the adaptive behaviour scale (ABS) parts I and II. We analysed differences in DAMES and ABS scores between groups with analyses of covariance or quantile regression in all patients who completed the 52 week assessment and had available follow-up data. This study is registered, number ISRCTN47562898. RESULTS We randomly allocated 88 patients to receive memantine (72 [82%] had DAMES data and 75 [85%] had ABS data at 52 weeks) and 85 to receive placebo (74 [87%] and 73 [86%]). Both groups declined in cognition and function but rates did not differ between groups for any outcomes. After adjustment for baseline score, there were non-significant differences between groups of -4·1 (95% CI -13·1 to 4·8) in DAMES scores, -8·5 (-20·1 to 3·1) in ABS I scores, and 2·0 (-7·2 to 11·3) in ABS II scores, all in favour of controls. 10 (11%) of 88 participants in the memantine group and six (7%) of 85 controls had serious adverse events (p=0·33). Five participants in the memantine group and four controls died from serious adverse events (p=0·77). CONCLUSIONS There is a striking absence of evidence about pharmacological treatment of cognitive impairment and dementia in people older than 40 years with Down's syndrome. Despite promising indications, memantine is not an effective treatment. Therapies that are effective for Alzheimer's disease are not necessarily effective in this group of patients. BACKGROUND Lundbeck.