BACKGROUND Knowledge of clinical transfusion practice should be improved to ensure that therapy is optimally effective, to avoid waste of resources and to ensure a safe supply of blood. METHODS All patients who received a transfusion of red blood cell concentrate at Ullevål University Hospital in two 14-day periods in 2003 were included. Diagnoses, haemoglobin values and intervention codes were recorded from the patient records for which consent to access had been given. Blood samples were taken from consenting survivors to be tested for blood group immunisation. RESULTS 348 patients were included. The median age was 62.8 years. They were given 1,162 concentrates in 471 transfusion episodes, of which 373 (79 %) consisted of one or two concentrates. As at 1 February 2009, 181 patients (52 %) were registered as having died. Access was possible to the records of 218 patients. The primary diagnosis was cancer for 76 patients (35 %), injuries for 36 (17 %) and cardiovascular disease for 34 (16 %). The transfusion was given to 89 (41 %) of patients in connection with a surgical intervention during the period covered by the patient records. A note about the transfusion was lacking in 46 (21 %) of the records. Transfusions were given to 52 patients whose haemoglobin concentration was above a threshold level of ≥ 8 g/100 ml (43 % of the patients). Blood group immunisation was found in one (3 %) of 38 survivors. CONCLUSIONS Red blood cell transfusion is most commonly given to elderly patients with chronic disorders and uncertain long-term prognoses. The clinical documentation is not infrequently incomplete. There is probably scope for a reduction in consumption if indications are based more on established scientific evidence and well-defined transfusion protocols. Blood group immunisation is not a frequent complication.