Alcohol is accepted as a cause of paroxysmal atrial fibrillation (PAF), supraventricular tachycardia and other cardiac arrhythmias1 but the effect is usually attributed to abuse rather than simple use.2 Abuse of alcohol – if this means the constant consumption in excess of 21-28 units a week for a man – is not always necessary in the initiation and prolongation of attacks of PAF, as my personal history indicates. Much lesser amounts may be responsible. I am over 70 years old and have been hypertensive for about 20 years and on various therapeutic regimens, mostly beta-blockers and thiazides at first, but more recently a combination of captopril, doxazosin and indapamide. My blood pressure, originally about 180/110, is now around 150/80 and my left ventricular T waves, at first flat or slightly inverted, are now normal. About 15 years ago I developed non-insulin dependent diabetes mellitus, now well controlled with a no-sugar diet, metformin 500 mg b.d. and chlorpropamide 100 mg a day. My ‘normal’ alcohol consumption was about 20-30 units per week as wine, with occasional excesses. About 12 years ago, after a reunion with a considerable intake of alcohol, I had my first attack of atrial fibrillation starting about 5 a.m. and lasting some four hours. Attacks thereafter were occasional at first, but gradually became more frequent, even with the accepted safe limit of alcohol intake. In 1993, I thought I was having some benefit from diuretics,3 and probably still do, but I am on indapamide for hypertension anyway (7.5 mg daily) and do not wish to experiment by leaving it off. About a year or so ago, attacks of atrial fibrillation were occurring three or four times a week and lasting up to 24 hours. Some attacks were followed by supraventricular tachycardia and one lasted four and a half days, being cut short only by sotalol. The prospect of a permanent supraventricular arrhythmia was then daunting. On the whole, antiarrhythmics were not much help. Continuous beta-blockers (atenolol or labetalol) and flecainide did not prevent attacks. In August 1996, I started experimenting by stopping all alcohol for some weeks, thereafter trying occasional amounts and assessing the effect. Two things quickly became apparent. Firstly, the effect of any alcohol consumption lasted for about five days; and secondly, the amount precipitating an attack did not need to be large – even one unit was enough. I am now virtually teetotal and limited to tasting wine and letting my friends drink it! Without alcohol I am not completely free from attacks but only have one every seven to 10 days or so; they last from one to three hours and never overnight. I could list about half a dozen precipitating factors. On the whole, the medical textbooks are gloomy about the prognosis of PAF, suggesting there is an inevitable natural progression to permanent atrial fibrillation. However, Paul Wood in the 1960s4 recognised that PAF was still paroxysmal four times out of five after an average of 10 years and that age was an important factor in its aetiology; but he did not even mention alcohol as a causative factor. It would be interesting to know how many people with PAF and taking an accepted safe amount of alcohol would benefit from either a drastic reduction in their intake or total abstinence. Most have probably not tried and are unlikely to be asked to do so by their medical attendants in view of the current belief that alcohol is safe in amounts of up to 28 units per week for a man and 21 units for a woman.