Atrial fibrillation (AF) is the most common arrhythmia requiring medical care, with a prevalence of almost 1% in the adult population. Particularly in the expanding elderly population, pharmacological therapy is and will continue to be the mainstay of AF therapy. Many currently used antiarrhythmic drugs have limited efficacy and cause cardiac and extracardiac toxicity. Thus, there is a continued need for development of new compounds with good efficacy and particularly with a favorable safety profile. Much emphasis is currently given to the development of so-called atrial-selective agents which target ion channels or proteins predominantly expressed in atrial myocardium. The rationale behind these compounds is to avoid unwanted effects on ionic currents on the ventricular site thus avoiding ventricular proarrhythmic effects. Alternatively, more conventional multichannel-blocking drugs are developed, for instance congeners of amiodarone. These molecules are designed to retain the electrophysiological efficacy of the mother compound while avoiding the extracardiac toxicity of this drug. The compounds which are far advanced in their clinical development are vernakalant ("atrial-selective") and dronedarone (multichannel-blocking). Preclinical and clinical findings of these substances are summarized.