In human, it has been shown that interactions between locomotor and respiratory patterns may lead to locomotor-respiratory couplings termed entrainment. In order to prove that this coupling is really an entrainment, we tried to show that it obeys one of the expected rules, i.e. that it evolves and is not present for all imposed locomotor frequencies. For that purpose, seventeen healthy volunteers were asked to run on a treadmill at 14 different locomotor rates (instead of 2 or 3 in previous works) for 40 s. All the subjects did not exhibit the same coupling and different relationships could be obtained: the most commonly observed was 2:1 (2 locomotor activities for a respiratory one) but other forms could appear (4:1 and even 5:2 or 3:2). When the coupling evolution was followed in the same subject, it did not appear for all locomotor frequencies but only for locomotor periods close to harmonics of respiratory ones (absolute coordination). On both sides of these values, it progressively evolved to relative coordination and to the lack of coordination. When two forms of absolute coordination were observed in a same subject, the phase relationships followed the rules of the entrainment. Compared to data obtained in quadrupeds, these results suggest that the entrainment of breathing frequency by the locomotor activity is due to central interactions between the respiratory and locomotor pattern generators and does not depend on a chemical regulation avoided here by short locomotor sequences.