Analysis of heart rate variability (HRV) has been used in studies of autonomic function and risk assessment in different patient groups such as in patients with diabetes mellitus, after myocardial infarction (MI) and other cardiovascular disease. Ectopic beats can, however, interfere with HRV analysis and give erroneous results. We have therefore studied the impact of ectopic beats on HRV analysis and the ability of a filter algorithm to correct this. Power spectral analysis of synthetic data with an increasing proportion of ectopic beats and 24-h Holter recordings from 98 healthy subjects and 93 post MI patients was done with and without digital filtering and interpolation of errors in the data stream. The analysis of HRV was seriously hampered by less than 1% of ectopic beats. A filter algorithm based on detection and linear interpolation of ectopic beats and other noise in the data stream corrected effectively for this in the synthetic data employed. In the healthy subjects and the post MI patients, filtering markedly reduced the extra variability related to non-normal beats. The software could automatically analyse over one hundred 24-h files in one batch. HRV analysis should include filtering for ectopic beats even with a small number of such beats. It is possible to make a fast analysis automatically even in huge clinical series, which makes it possible to use the method both clinically and in epidemiological studies.