Despite multiple efforts to increase cyclists' awareness of traffic regulations, the way in which rule knowledge translates into safer behaviours remains insufficiently understood, as do the conditions under which these effects are stronger (e.g., Volgemute et al., 2024). This study examines whether risk perception mediates the relationship between cyclists' knowledge of traffic rules and their self-reported behaviours, and whether previous crash involvement moderates these effects. Data were drawn from 6845 cyclists from 19 countries (aged 18-83 years; M = 36.88, SD = 14.65), a dataset published by Useche et al. (2024). It includes the Cycling Behaviour Questionnaire (CBQ; measuring violations, errors, and positive behaviours) and the Risk Perception and Regulation Scale (RPRS; assessing cyclists' risk perception and their self-reported knowledge of traffic norms). A moderated mediation model was applied, controlling for age, gender, and country. Results suggest that cyclists with greater rule knowledge report higher risk perception, with this association stronger among those who had experienced a crash. Rule knowledge was negatively related to violations only for previously crash-involved cyclists, whereas higher risk perception was consistently associated with fewer violations, slightly more strongly among those having suffered cycling crashes. Both rule knowledge and risk perception were associated with fewer errors and more positive behaviours. The effects on errors are more pronounced for those who have experienced at least one crash, whereas no interaction effects were observed on positive behaviours. The benefits of rule knowledge are more pronounced and direct on errors (negatively) and positive behaviours (positively). Overall, findings position risk perception as a proximal mechanism linking rule knowledge to actual riding behaviour and suggest pairing rule training with hazard perception to maximise safety gains.