The analysis of heart rate variability is supposed to be a marker of autonomic cardiac activity and is used for risk stratification of post-infarction patients. Analysis of heart rate variability in the frequency domain may permit a differentiation of vagal and sympathetic control; for such analyses only short time intervals characterized by a steady-state autonomic balance can be used. Yet, it is unclear whether single determinations of heart rate variability indices derived from short time intervals yield reproducible results. Therefore, the reproducibility of heart rate variability indices was studied with weekly measurements in 10 healthy volunteers under the following defined conditions: 13 min supine rest, 10 min standing, 13 min sitting, and 15 min cycle ergometry followed by a 14 min recovery period. Heart rate variability was determined in the frequency domain (fast Fourier transformation) and in the time domain. The reproducibility was estimated by the coefficient of variation (CV). Additionally, the reproducibility of heart rate, blood pressure, and the expiratory-inspiratory ratio of heart rate was determined. The reproducibility of the frequency domain indices (36.6-74.9% CV) and of the time domain indices (19.6-32.8% CV) was considerably worse than that of heart rate (5.2-8.2% CV), blood pressure (5.1-8.2% CV) and the expiratory-inspiratory ratio of heart rate (4.6% CV). The reproducibility of heart rate variability indices was not improved by orthostatic or ergometric challenge. This poor reproducibility does not permit a reliable interpretation of heart rate variability on the basis of single measurements in healthy volunteers. Given the wide range and scatter of the measured parameters, the diagnostic and prognostic value of heart rate variability indices derived from short recording periods appears questionable.