Nalmefene was the first treatment approved by the European Medicines Agency for reducing alcohol consumption in adult patients with alcohol dependence. It is often presented as a paradigm shift in therapeutics, but major issues limit the interpretation of the evidence supporting its use. The randomised trials submitted provided no evidence of harm reduction, the differences on consumption outcomes were of questionable clinical relevance, the target population was defined a posteriori and the drug was compared to a placebo although naltrexone was already used off-label. No post-approval randomised study is currently designed to clearly address these issues. In addition, nalmefene trials have been uncritically cited, even in guidelines. This experience reveals weaknesses in drug evaluations in alcohol dependence, which call for changes. We propose to dispense with alcohol consumption as a surrogate outcome, to consider comparative effectiveness issues, and to recommend randomised post-approval studies in case of controversial approval.