David Taylor is right that there is excessive polypharmacy in routine practice. ^{Reference Taylor1} However, he does not examine or comment upon one of the root causes, British National Formulary (BNF) limits. Many clinicians seem to believe they are acting in the patient's interest by prescribing two compounds at close to the BNF maximum rather than one above this mark. As a clinician it is commonplace to come across patients who respond well to sub-BNF doses as well as those who are untouched by a drug at the BNF maximum dose. In the case of antipsychotic drugs, Agid et al ^{Reference Agid, Mamo, Ginovart, Vitcu, Wilson and Zipursky2} have once again demonstrated that response to these drugs is related to the measured blockade of striatal receptors. As I suggested in my paper 12 years ago, ^{Reference Searle3} this allows the clinician to quickly and accurately judge the sensitivity of an individual patient to antipsychotic treatment by increasing the dose rapidly to the point at which extrapyramidal side-effects are just discernible - and then waiting for a response. Following this threshold dosage scheme has led me to occasionally use a much wider range of doses than the BNF limits allow. For example, I have prescribed risperidone in schizophrenia with good effect at as little as 0.5 mg per day and as much as 32 mg per day, a 64-fold dose range. Although those who practise acute adult psychiatry often observe that patients with severe psychosis may be dramatically medication-resistant, unless they have used threshold dosing they do not know that the sensitivity of the patient to antipsychotic medication increases as their mental state improves, allowing a reduction in dose with maintained efficacy. It is worth remembering that BNF limits are usually established in accessible and responsive out-patient populations with moderate symptoms. Practising clinicians treat many patients who do not come from this population and may find themselves with a difficult choice: polypharmacy or prescription outside BNF limits.