summary Previous studies of blood use have used different methods to obtain and classify transfusion indications. Before undertaking a national study of transfusion recipients, a pilot study was performed over 2 months at two teaching and two district general hospitals to match information from hospital transfusion laboratories with clinical coding data from the hospital’s Patients Administration System to determine the indication for transfusion in 2468 recipients. Data analysis revealed major limitations in the conventional use of primary diagnostic International Statistical Classification of Disease and Related Health Problems 10th Revision (ICD-10) or procedure Office of Population, Censuses and Surveys – Classification of Surgical Operations and Procedures – 4th Revision (OPCS-4) codes alone in allocating transfusion indications. A novel algorithm was developed, using both types of code, to select the probable indication for transfusion for each patient. A primary OPCS-4 code was selected for recipients transfused in relation to surgery (43%) and either the primary (36%) or the secondary (12%) ICD-10 code was chosen for recipients transfused for medical reasons. The remaining patients were unclassified. Selected codes were then collated into Epidemiology and Survival of Transfusion Recipients (EASTR) casemix groups (E-CMGs). The most frequent E-CMGs were haematology (15% of recipients), musculoskeletal (14%), digestive system (12%) and cardiac (10%). The haematology E-CMG includes patients with malignant and non-malignant blood disorders and recipients transfused for anaemia where no cause was listed. Recipients undergoing hip and knee replacement and coronary artery bypass grafting are within the musculoskeletal and cardiac E-CMGs. The digestive E-CMG includes recipients transfused for gastrointestinal (GI) bleeds and those undergoing GI surgery. This methodology provides a more useful means of establishing the probable indication for transfusion and arranging recipients into clinically relevant groups.