Atrial fibrillation is a common arrhythmia in an intensive care unit. We performed a prospective observational study over a period of three months, to study the incidence, risk factors and outcome of patients who develop atrial fibrillation in a multidisciplinary intensive care unit. All patients above the age of 50 years were eligible. Exclusion criteria were: cardiac or oesophageal surgery during current hospitalisation, atrial fibrillation at admission, implanted pacemaker and expected intensive care unit stay of less than 24 hours. Sixty-one patients were included in the study. Eighteen patients (29.5%, confidence interval 18-40) developed atrial fibrillation. Incidence of atrial fibrillation was 4.02 episodes per 100 patient days. Patients who developed atrial fibrillation had higher age (71.3 years vs. 63.2 years, P=0.001), severity of illness (APACHE II 25.4 vs. 20.0, P=0.005) and sepsis at admission (9/18 vs. 9/43, P=0.01). They also had higher in-hospital mortality (Risk ratio 2.7, 95% confidence interval 1.3-5.4). Standardised mortality ratio was higher in patients who developed atrial fibrillation (1.08 vs. 0.63). Patients who developed atrial fibrillation required a longer period of mechanical ventilation and inotropic support. Multivariate logistic regression analysis showed age >75 years, APACHE II score >20 and sepsis at admission were independent predictors for development of atrial fibrillation in critically ill patients. Although atrial fibrillation by itself is unlikely to be the cause of higher mortality, it is likely to be a marker for increased mortality and resource utilisation in the intensive care unit.