The arrhythmogenic effects of exercise tests are often unpredictable, sometimes paradoxical, and usually poorly reproducible. This explains why they are rather neglected by rhythmologists in studying the mechanisms of arrhythmias and the effects of antiarrhythmic drugs and why their prognostic value is a matter of controversy. At the same time the induced, supposedly ischaemia-related arrhythmias are usually not prevented by revascularization procedures. All these discrepancies are probably explained by the complex interactions between the time-limited stimulation of the sympathetic drive and the arrhythmia mechanisms. During the exercise test the sympathetic drive is increased via the neurogenic and humoral routes, and the post-exercise period is characterized by a vagally-induced deceleration of the heart rate which intervenes in the setting of a persistent high level of sympathetic activity. The arrhythmia components formed by potential triggers (isolated premature beats) and substrates (repetitive activity) do not react in the same way to those different phases. Extrasystoles are mainly rate-dependent, so that they often tend to disappear as the sinus rate accelerates. The repetitive ventricular activity directly depends on the sympathetic drive, but necessitates the presence of the trigger to become manifest. Conceivably, many different situations may result from the complex interplay between these distinct parameters, so that the apparent clinical effect of exercise may be either arrhythmogenic or antiarrhythmic in apparently identical diseases, or even in the same patients.