<p>This study aimed to provide a multiangled comparison of the impact of real and virtual setups on the behaviour and perception of electric scooter (e-scooter) riding to examine the external validity of the virtual environments in the micro-mobility context. For this purpose, a within-subject design experiment was conducted to collect data on the behaviour of e-scooter riders. Furthermore, self-reported data on mental and physical demand as well as physiological data in the form of heart rate measurements and electroencephalography (EEG) were recorded to provide an additional insight into the behavioural results. The analysis was based on the multinomial logit model (MNL) for the behavioural data, ordered logit models for self-reported measures. Parametric and non-parametric tests were performed to capture the differences in physiological data. The results of the behavioural data showed significant differences in braking and acceleration patterns between virtual and real scenarios, which undermined the external validity of virtual environments in the current context. Further, self-reported measures painted a mixed picture when looked at jointly with biometric measures, where the questionnaires indicated that both setups were indifferent with respect to mental demand, while the physiological data suggested that virtual scenarios were more mentally engaging for the e-scooter riders.</p>