The first point raised is that the delay to clozapine initiation may not be a true reflection of the actual delay because patients may have been offered clozapine but refused it. This, of course, depends on what delay you are interested in. In our study we used the delay from the point at which treatment guidelines recommend a patient should start clozapine. ^{Reference Howes, Vergunst, Gee, McGuire, Kapur and Taylor1} In our view this is the key, clinically relevant, delay. However, Sharma & Grover are right in suggesting that this delay does not necessarily mean that clinicians have delayed offering clozapine, although if this were the case it implies that it has taken on average 4 years for patients to agree to start clozapine. In practice it seems likely that there are a number of patient, clinician and service factors that may underlie the delay we observed in our study. Understanding these will be important if delays are to be reduced in the future. The availability of biomarkers for treatment resistance, as indicated by a recent study, ^{Reference Demjaha, Murray, McGuire, Kapur and Howes2} could also contribute to identifying treatment-resistant patients earlier. Sharma & Grover also rightly raise the issue that duration of untreated psychosis was not assessed in our study. Consequently, we cannot exclude the possibility that the duration of illness was in fact longer in our sample and thus that the delay to effective treatment was in fact longer.