Recent research has shown that increased injury risk for supervisees during sibling supervision is in part due to supervision practices of older siblings. The current study was the first to test an online training program for sibling supervisors. A randomized controlled trial design was used, with older siblings randomly assigned to either an intervention (n = 29) or control (n = 26) group. Before and after either the intervention or waitlist period, older siblings completed measures of supervision knowledge and their supervision behaviours were directly observed in the laboratory with their younger siblings. Results revealed that, compared to the control group, older siblings who completed Safe Sibs showed significant improvement in several areas of supervision knowledge (child development, what constitutes effective supervision practices, injury beliefs, intervention-specific knowledge) and a non-significant trend towards improvement in knowledge of effective child management strategies. Importantly, there was also statistically significant improvement in supervision behaviour, namely- frequency of proactive safety behaviours to prevent supervisee access to injury hazards. Implications for sibling supervision and injury risk, as well as limitations and directions for future research, are discussed.