This editorial on the use of intramuscular clozapine^{Reference Bhattacherjee, Young, Ahuja and Allen1} has the potential to mislead readers. The authors question the efficacy of intramuscular clozapine on the grounds that it is not always given when prescribed^{Reference Casetta, Oloyede, Whiskey, Taylor, Gaughran and Shergill2,Reference Henry, Massey, Morgan, Deeks, Macfarlane and Holmes3} and go on to recommend the use of intramuscular haloperidol or olanzapine as alternatives to intramuscular clozapine. Most patients who respond to clozapine are willing to continue taking it once their insight has improved but may be initially reluctant while acutely unwell. In many instances, a short period of assertive treatment is justified in order to establish the patient on an effective long-term treatment which they will ultimately accept, and this is where intramuscular clozapine is useful. All the patients in the study had declined to take clozapine prior to being prescribed intramuscular clozapine. Once prescribed intramuscular clozapine, all were again encouraged to accept oral treatment as an alternative to intramuscular. As the data show, around half then accepted oral treatment without a single administration of intramuscular clozapine but would not have done so had intramuscular clozapine not been prescribed. Intramuscular forms of haloperidol and olanzapine may have the advantage of being licensed products (although there is no UK-licensed intramuscular olanzapine at the moment), but their use in treatment-resistant patients is ethically unsupportable given the near certainty that they will be ineffective as antipsychotics in this patient group. In Kane's landmark study of clozapine,^{Reference Kane, Honigfeld, Singer and Meltzer4} 305 enrolled patients were initially treated with haloperidol at an average dose of 61 mg/day. Fewer than 2% of patients responded, and there was no mean change in symptom score for this cohort as a whole. In the study proper, 30% of these patients responded to clozapine within 6 weeks. Likewise, in a smaller study, olanzapine 25 mg/day was associated with response in only 5% of a treatment-resistant group, and 41% of the same patients subsequently responded to clozapine.^{Reference Conley, Tamminga, Kelly and Richardson5} Some studies have shown benefit for non-clozapine antipsychotics in resistant patients, but these trials are methodologically flawed and subject to funder bias.^{Reference Taylor6} Most clinicians accept that clozapine is uniquely effective in refractory schizophrenia. We agree with the authors that intramuscular clozapine might have limited potential as an ad hoc intervention to prevent gaps in treatment, but not primarily because of the time this would take to arrange. The main problem with using intramuscular clozapine for those on higher maintenance doses is that the maximum oral equivalent dose to one 4 mL injection is 200 mg, and large variation in clozapine dosages can be dangerous. Rather, intramuscular clozapine is most useful as part of a pre-discussed and well-planned multidisciplinary team initiation regimen. The editorial's authors draw the reader's attention to the risks associated with inadvertently administering an overdose of intramuscular clozapine to a clozapine-naïve patient. This is equally important for oral clozapine, of course, and the two formulations have a similar duration of action. Any use of unlicensed medication carries risks and should only be done with appropriate safeguards, in appropriate settings, and following a thorough appraisal of risks versus benefits, involving the patient and their carers wherever possible. Where the benefits outweigh the risks, intramuscular clozapine can be the only route to being successfully started on this uniquely effective drug. As Casetta and colleagues showed,^{Reference Casetta, Oloyede, Whiskey, Taylor, Gaughran and Shergill2} the great majority of patients who commenced clozapine responded well and continued to take it. Without intramuscular clozapine, such patients would have remained ineffectively treated.