Introduction: As technology progresses, individuals will be increasingly expected to solve digital tasks. At the same time, many worry that a high use of connected devices will reduce young people's ability to perform with accuracy long cognitively challenging tasks online. Methods: We examine whether 15-year-old students' ability to accurately solve cognitively challenging digital tasks--and to maintain accuracy throughout the 2-h PISA low-stakes--reflects their frequency of internet use. We do so using data from 153,603 students from 27 countries who participated in the 2018 Programme for International Student Assessment (PISA). Results: Compared to students with moderate levels of internet use, students who use the internet a lot and those who never use it (or use it very little), have lower baseline levels of accuracy in all key assessment domains (reading, mathematics and science). By contrast, students' use of the internet is not associated with how much accuracy declines over the 2-h assessment when students are required to solve mathematics and science tasks. In reading, students who use the internet a lot have lower declines in accuracy over the course of the 2-h assessment compared to students with medium levels of internet use who, in turn, have lower declines than students with low levels of internet use. Internet use is not associated with how carefully students respond to questions in the background questionnaire. Conclusions: Worries about internet use reducing young people's persistence appear unfounded. At the same time high levels of internet use are associated with low baseline accuracy.