It is not trivial to understand the interactions between heart rate variability (HRV), activity of the autonomic nervous system (ANS) and exercise training. In this way, heart rate (HR) signals were analyzed. HR was recorded during a race of 82 km from two groups of runners performing different training regimes, Active and Elite. Several indexes from time-domain and frequency-domain analysis, time-frequency representation (TFR) and auto-mutual-information function (AMIF) were calculated on RR series in order to describe their dynamicity. TFR and AMIF indexes presented statistical significant differences when comparing the 1st and the 6th hour of the race (p<0.002). LF/HF showed an increasing tendency in Active runners, and a decreasing tendency in Elite runners. Extremely low values of RR standard deviation were found in Elite runners.