The N-methyl-D-aspartate receptor antagonist, memantine, is licensed for the treatment of moderate to severe Alzheimer’s disease (AD). Memantine is administered both as a monotherapy and as an add-on therapy in patients already receiving acetylcholinesterase inhibitors. Several meta-analyses have been published that examine the efficacy of memantine in the treatment of AD, based on clinical trial data. However, different disease severities and concomitant medication use in the trial populations means that synthesis of this data is challenging with numerous methodological decisions required. The main objectives of this study were to review the methodologies of different meta-analyses, assess the impact of specific methodological approaches on efficacy results, and to help interpret previous meta-analyses results concerning the efficacy of memantine in moderate to severe stages of AD. The methodologies of five meta-analyses were reviewed in terms of the included trials, combination of data, choice of outcome, and analysis methods. Results were extracted and compared in line with the methodological approach taken. The most robust results were observed on cognition, activities of daily living, and overall assessment, where memantine showed a consistent benefit over placebo. The benefit of memantine on behavioral symptoms was also demonstrated, but results were more heterogeneous. Variability could not be explained by baseline severity and concomitant treatment alone. It is stressed that interpretation of meta-analysis results must be considered within the context of the methodological approach. Overall, results from individual clinical trials and from meta-analyses demonstrate that memantine represents a valuable treatment option in AD.