The analysis of heart rate variability (HRV) is a powerful tool in the study of the autonomic control of the heart. While circadian HRV rhythms have widely been characterized by traditional spectral measures, ultradian oscillations are not commonly investigated. In this study the identification of HRV ultradian rhythms is assigned to a quantitative measure characterizing the fractal-like behavior of the time series: the fractal dimension (FD). In order to assess ultradian regulation in Chronic Obstructive Pulmonary Disease (COPD) 24-h Holter ECG recordings of 52 COPD and 10 normal (healthy) subjects were analyzed. The FD was calculated by Higuchi\'s algorithm both during daytime and nighttime to highlight possible wake-sleep states differences. All subjects showed a similar common rhythm (0.06mHz) that persists with generally higher amplitude during night-time. A further rhythm becomes predominant in normal subjects in the day-to-night transition (0.15mHz), probably under the influence of the REM/non-REM ultradian sleep cycle. A very large difference between night and day amplitudes of this rhythm and of the next one (at about 0.22mHz) characterize the HRV fractal dimension of the Normal in respect of COPD. In conclusion, the FD could be used as a marker of ultradian cardiac autonomic regulation providing new insights into autonomic physiology of normal and COPD patients.