OBJECTIVE--To assess current strategies used to investigate and manage acute atrial fibrillation in hospital. DESIGN--Prospective survey of all acute admissions over 6 months. SETTING--District general hospital serving a population of 230,000 in north east Glasgow. SUBJECTS--2686 patients admitted as emergency cases over 6 months. RESULTS--Of the 2686 patients, 170 (age range 38-95, mean (SD) 73.5 (10.6) years; 70 men (41%) and 100 women (59%)) were admitted with atrial fibrillation. The principal underlying medical conditions were ischaemic heart disease in 79 (46.5%), rheumatic heart disease in 26 (15.3%), and thyroid disease in six (3.5%). Cardiac failure was present on admission in 61 (36%), cerebrovascular events in 23 (14%), and myocardial infarction in 17 (10%). Of those with a history of atrial fibrillation (102 (60%) including 10 with paroxysmal atrial fibrillation) treatment on admission included digoxin in 71 (70%), warfarin in 20 (20%), and aspirin in 17 (17%); the aspirin was predominantly given for concomitant vascular disease. The mean (SD) inpatient stay was 16 days (19.7) (range 1-154) largely due to the patients with stroke. Thyroid function tests were performed in only 63% and echocardiography in 33%. Overall, the rate of introduction of anticoagulation (seven patients) and attempted cardioversion (21 patient: 19 pharmacological and two electrical) was surprisingly low. Only 49 patients (34% of those not on warfarin) had contraindications to anticoagulation: these included peptic ulcer or gastrointestinal bleeding in 18 (12%), dementia in eight (6%), chronic renal failure or dialysis in eight (6%), and alcohol excess in four (3%). CONCLUSION--Standard investigations were inadequately used in patients with atrial fibrillation and there was a reluctance to perform cardioversion or to start anticoagulant treatment.