The aim of this study was to examine costs of charted drugs by using three data capture techniques. Data were manually extracted retrospectively from drug charts for a stratified sample of 200 inpatient stays at each of two London hospitals — a teaching hospital (St George's) and a district general hospital (Kingston). Specialties where costing was not feasible were excluded. These were: intensive care, oncology, dermatology and paediatrics. Three costs were calculated: nonstock drugs administered (the capability of traditional pharmacy computer systems), drugs prescribed (the data captured by a basic computerised prescribing system) and drugs administered (the data captured by a drug administration recording system). Pharmaceuticals not prescribed on the drug chart, such as anaesthetics, contrast media and disinfectants, were not included.At St George's 2,739 and at Kingston 2,843 prescription entries were recorded, amounting to costs of £2,940 and £2,327, respectively, for drugs administered. A range of £0.01 to £171 per patient (median £4.67) at St George's and £0.00 to £122.66 (median £2.88) at Kingston demonstrates a skewed distribution, with most patients incurring low costs. Non‐stock drugs accounted for 21 per cent of the cost of drugs administered at St George's and 15 per cent at Kingston. Twenty‐four patients incurred costs above £50, their non‐stock drugs being a mean of 21 per cent of their total costs. The cost of drugs administered at both hospitals was found to be 95 per cent of the cost of drugs prescribed.Computerised prescribing and drug administration recording systems would significantly improve cost allocation. However, it is unlikely that the expense could be justified on the basis of the low value of drugs received