Without treatment, about 60% of atrial arrhythmia patients suffer a relapse within 3 months and 70% within one year. Antiarrhythmic treatment intended to reduce this percentage is therefore justified, on condition that it is well tolerated. Several preliminary questions have to be settled before this medical prophylaxis: 1) Justification of antiarrhythmic treatment (sometimes pointless to deal with very occasional episodes); 2) Treatment of the underlying heart disease (valve disease, cardiothyrotoxicosis, etc.) or promoting factors (potassium depletion etc.); 3) Accurate assessment of any associated conduction abnormalities, which may constitute a contraindication to antiarrhythmic treatment (WPW syndrome in the case of verapamil and the digitalis-like drugs) or require additional treatment (pacemaker); 4) Definition of the mechanism (vagal or sympathotonic) inducing arrhythmia; 5) Evaluation of the hemodynamic parameters of the underlying heart disease (size of the atria, ventricular function, coronary or valvular lesions) which may limit the efficacy of the treatment. Once these parameters have been identified, the primary treatment should be type la or lb antiarrhythmics, which have been shown to be effective, despite the fact that they are not without arrhythmic risks (the Ib antiarrhythmics are less effective and have a poor safety profile). The beta-blockers have preferential indications (hypersympatheticotonia, hyperthyroidism, hypertrophic myocardiopathy, mitral prolapse, angina etc.) and can be replaced by verapamil or bepridil if there are non-cardiac contraindications (ulcers, asthma, diabetes). Amiodarone is extremely effective, but its poor extracardiac safety restricts its long-term use. Complementary treatments (digitalis-like, anticoagulants or anti-PAF and cardiostimulant drugs) should be added if necessary. Recurrences (to be confirmed by ECG or Holter) should lead to rigorous confirmation of therapeutic compliance and observance of simple hygienic and dietary measures (no excessive exertion, elimination of stimulants etc.). With strict clinical and ECG monitoring, it would then be possible either to increase the dose levels (accompanied by plasma determinations if possible) or to switch to a treatment with more effective, but more aggressive drugs (amiodarone, flecainide) or to use drug associations (la and lb, la and II etc.). Repeated failure of such attempts should lead to a non-medical approach to treatment.