The rider's comfort is an essential parameter in the design of a bike. A cyclist can often spend more than six hours on their bikes during a race and accumulate over 3000 km during competition. However, recent studies demonstrated that exposure to vibration for only 7 minutes on a cobblestone road can cause detrimental effects. The aim of this work is to apply operational modal analysis (OMA) method to study the dynamic behaviour of a bike-rider system. The experiments is conducted in two stages: (i) a voluntary subject (175 cm, 65 kg) carried out 10 tests on a cobblestone road at a speed of 20 km/h to study the repeatability of the method, and (ii) two volunteers (175 cm, 65 kg and 190 cm, 100 kg) performed a test on a cobblestone road at speeds of 10 and 20 km/h. For these two stages, the bike used has an aluminium frame and a carbon fork. 14 piezoelectric sensors are mounted on the bike for the OMA. The analysis was performed using the Modal Analysis 5.2 software (OROS, Grenoble, France) for both stages of the protocol. A frequency domain technique is applied to extract the modal parameters of the bicycle structure. The modes are identified in the range of 5-50 Hz. It turns out that the OMA gives repeatable results for the modal frequencies (coefficient of variation CV = 4.2%), but not for the modal damping (CV = 75.3%) and mode shapes (modal vectors accuracy = 12%). It is also shown that the speed and mass of subjects significantly influences the estimated modal frequency (relative deviation of 37.0% and 17.2% respectively).