Heart rate variability (HRV) is considered to indicate cardiac autonomic reflex activity. The reproducibility of HRV determinations from repeated Holter recordings has not been vigorously evaluated. This represents, however, an important question since HRV determinations are increasingly used to examine disease-related changes in the autonomic tone or the effects of therapeutic interventions. Thus, the present study examined the reproducibility of 4 commonly used time-domain parameters (SDNN, SDANN, rMSSD, pNN50) and the spectral components of HRV in 17 healthy volunteers and in patients with coronary artery disease (n = 9) or hypertensive heart disease (n = 9). In all subjects, 2 Holter recordings were obtained 7 days apart. In both, healthy volunteers and in patients, overall reproducibility of HRV parameters was good exhibiting correlation coefficients between 0.60 and 0.98. However, in individual subjects particularly in those without heart disease a higher degree of variability of HRV measurements was observed. In those patients with markedly reduced HRV, reproducibility was best. It is concluded that repeated HRV determinations may be utilized to examine the effects of underlying disease progress or of therapeutic interventions on cardiac autonomic tone.