This study utilised 2015 PISA data from the United Kingdom and Hong Kong to examine the relations between outside school ICT use for learning and science performance through a mediation model. The results uncovered three key findings: Firstly, ICT use for learning exhibited negative total effects on science performance in the United Kingdom, whereas the effects were non-significant in Hong Kong. Secondly, ICT use for learning consistently showed positive indirect effects on science performance through the mediation of self-efficacy, enjoyment and interest. However, no indirect effects were observed through utility value in either of the sampled countries/regions. Notably, ICT use for learning showed positive indirect effects through the mediation of epistemological beliefs in Hong Kong but not in the United Kingdom. Thirdly, ICT use for learning displayed negative unexplained (i.e., direct) effects on science performance, suggesting that potentially negative mediators have not been fully identified. Future research shall focus on exploring potential mediations that may cause negative indirect effects to develop more targeted interventions.