The study looked at palpitations in relation to the prevalence of arrhythmia, as assessed by 24-h ambulatory electrocardiography (ECG) in a population sample. The subjects were randomly drawn from among those involved in a cardiovascular survey. Forty-three of those who answered 'Yes' and 54 of those who answered 'No' (84% of those eligible) to the following question, participated: 'Have you observed sudden changes in your heart rate or heart rhythm during the preceding year?' In both groups mean age was 49 years and 58% were men. There was no relationship between recorded arrhythmia and perceived palpitations during monitoring. The prevalence of at least one arrhythmic episode (ventricular or supraventricular arrhythmia or pauses > = 1.5 s) was significantly higher in those who had perceived palpitations during the previous year (98%) than in those who had not (74%) (P < 0.0014). Through a simple question about palpitations during the preceding year we were able to identify significantly a population with true arrhythmias. However, the question could not be used to define a population entirely without arrhythmia. The high prevalence of arrhythmia in subjects without reported palpitations indicates that it is a normal finding which alone should not demand further clinical investigations.