Nonlinear analysis of HRV, particularly the fractal dimension and 1/f beta parameters, has been recognized to provide valuable information in the prognostic classification of cardiac patients. The reproducibility of these methods, however, is not known. In this study we addressed these issues in a population of chronic heart failure (CHF) patients. We analyzed 3 ECG Holter recordings from 22 clinically stable CHF patients by 1/f log scaling of power spectrum and the fractal dimension of the HRV curve, assessing the reproducibility of each index. Our results show that Fractal dimension parameter exhibits similar reproducibility and a higher reliability than both 1/f and spectral parameters. Moreover, we found that fractal dimension index is quite correlated with 1/f slope and describe possibly complementary aspects of heart rate dynamics.