1. The fractal scaling of heart rate variability, gauged by the correlation dimension (CD), is hypothesized to be characterized by a time structure (chronome), which in health shows differences as a function of gender and age. 2. From 24 h Holter records of 44 clinically healthy male subjects in four age groups (5-10, 20-25, 40-45 and 60-65 years; n = 11 in each group), 500 s sections at 4 h intervals for 24 h were analysed for smoothed R-R intervals sampled at 4 Hz. Using an algorithm modified from Grassberger and Procaccia (Physica D 1983; 9: 189-208), the correlation integral was estimated for embedding dimensions from 1 to 20 with a 1.0 s time lag for each section. Nightly (02.00 hours-06.00 hours) ECG records were similarly analysed in 72 additional clinically healthy subjects of both genders, 5-70 years of age. The single cosinor assessed the circadian characteristics; one- and two-way analyses of variance and linear regression were used to examine changes as a function of gender and age. 3. The 24 h average of CD is largest in the 20-25-year-old men and decreases with age there-after (P < 0.05). These changes apply in particular to the nightly CD values, which are higher in female than in male subjects (P < 0.001). Increasing age is associated with a decrease in the amplitude and an advance in the phase of the circadian rhythm in CD (P < 0.05). 4. A chaotic end-point from fractal scaling, yielding a non-linear index, such as the correlation integral, undergoes a circadian rhythm and changes with gender and age. This assessment in the chronome represents an added diagnostic tool in cardiology, and provides new end-points for the study of coherence among internal variables of autonomic mechanisms and of influences by external environmental variables upon them.