With the world becoming increasingly digitalized, determining the relationship between the use of Information and Communication Technologies (ICT) and educational outcomes takes on special relevance for guiding educational policy decisions in a reasoned way. The objective of this research is to explore the relationship between different types of ICT use at school and at home, students' attitudes towards ICT, and academic performance, as well as to see if these associations differ according to the level of performance of the students. For this purpose, we apply multilevel regression models and quantile regression models with data from the Programme for International Student assessment survey (PISA) 2015. The results show differences in the sign of the association according to the ICT variable analysed. The positive association between ICT use and academic performance is related with its use for entertainment at home, the students' interest in ICT and an earlier starting age for using ICT. Results also show that higher availability of computers per student in the schools is associated with higher levels of academic performance. However, the use of ICT for schoolwork at home, the use of ICT by students in schools and the importance of ICT as a topic in social interaction for students, are factors negatively associated with academic performance. Finally, our quantile regressions by ability levels suggest that certain types of ICT use appear to be particularly associated with the academic performance of students in the lowest percentiles of performance. These results highlight the need to consider and include ICT as an input into the educational production function. Moreover, public policies should consider the interaction between ICT and the educational performance of underperforming students.