Atrial fibrillation occurs frequently in medical intensive care unit patients. Most intensivists tend to treat this rhythm disorder because they believe it is detrimental. Whether atrial fibrillation contributes to morbidity and/or mortality and whether atrial fibrillation is an epiphenomenon of severe disease, however, are not clear. As a consequence, it is unknown whether treatment of the arrhythmia affects the outcome. Furthermore, if treatment is deemed necessary, it is not known what the best treatment is. We developed a treatment protocol by searching for the best evidence. Because studies in medical intensive care unit patients are scarce, the evidence comes mainly from extrapolation of data derived from other patient groups. We propose a treatment strategy with magnesium infusion followed by amiodarone in case of failure. Although this strategy seems to be effective in both rhythm control and rate control, the mortality remained high. A randomised controlled trial in medical intensive care unit patients with placebo treatment in the control arm is therefore still defendable.