I read with interest and some astonishment the work of Liebrenz et al. [1], which suggested benzodiazepine ‘substitution’, preferably with long-acting benzodiazepine (BZD), for long-term users. There are no empirical data to support the authors' view, so the discussion about the proposed procedure is academic; however, a few critical issues are apparent. Treatment of long-term benzodiazepine users is undoubtedly a difficult and often disappointing task, as is addiction medicine per se. A long-term study of patients with ‘complicated’ withdrawal showed that only 25% remained BZD free [2]. There are a number of reasons for discontinuing BZDs, including improvement of psychomotor and cognitive performance and also the risk of personality changes often seen in long-term users (something not mentioned by the Liebrenz paper). The empirical basis considering the optimal pharmacological strategy for withdrawing BZDs is limited, but the clearest strategy is to taper the medication [3,4]. A Cochrane analysis found only carbamazepine to be helpful with modest benefits in reducing withdrawal severity [5]. In addition, short half-life BZDs are associated with higher dropout rates, but switching from short-life BZDs to long half-life BZDs before gradual taper withdrawal did not receive much support from the Cochrane review. Nevertheless, Lader et al. [4] suggested that the substitution of diazepam for another BZD can be helpful, ‘at least logistically’, because diazepam is available in a liquid formulation. There are also a number of studies on optimal psychotherapeutic interventions [2,6,7], with cognitive–behavioural interventions probably having the best evidence. Terminology is sometimes important to avoid misunderstandings. As a non-native English speaker I wonder whether the term ‘substitution’ is really adequate for the suggested procedure. Maintenance may be the better term. The real question is whether or not to discontinue BZDs, which is difficult to decide based upon patient characteristics. Both high-dosage BZDs [2] and intensive craving for BZDs [8] are associated with negative outcome. No one would argue that replacing vodka with beer is a form of substitution treatment, and it is difficult to see why this should be different with respect to drugs. Taking into account the small number of clinical trials in this area, the therapeutic pessimism of the Liebrenz paper and the call for ‘agonist’ substitution seems to be a little premature. Patients with BZD dependence are undertreated compared to patients with alcohol or other substance use dependence [9], so the primary research focus might be to get more patients into treatment, and into treatment earlier, and to develop better evidence-based tactics rather than following an approach of unconditional surrender by simply replacing one BZD with another. Finally, some pharmacological comments. As outlined above, the recent Cochrane analysis does not support the authors' view on half-lives as being the essential point—although it is easy to see that a long-acting compound may be easier to handle compared with short-acting ones. Longer-acting drugs may have a higher risk of accumulating in the body and causing hangover. Many BZDs have pharmacologically active metabolites and may therefore also not be optimal for ‘substitution’ treatment. It may make sense to study other compounds such as BZD receptor modulators, as suggested by Denis et al. [5], or other psychoactive drugs, especially as most patients with BZD dependence have psychiatric disorders, rather than continuing travelling on a ship that should have been abandoned a long time ago. None.