Background: Digital texts are progressively becoming the medium of learning for students, but research has indicated that students tend to process information more superficially while reading on screen. It is therefore relevant to examine what strategies can support digital text comprehension. Objectives: This study aimed to investigate the effects of highlighting - both learner generated and experimenter provided - when reading digitally. Methods: University students (N = 170) were randomly assigned to the condition of learner-generated highlighting, experimenter-provided highlighting, or control. Reading outcomes were measured as literal and inferential text comprehension, transfer of knowledge, and metacognitive calibration of comprehension performance at immediate and delayed post-tests. Individual differences in prior knowledge, cognitive reflection, and reading self-efficacy were taken into account. The quality of the information highlighted by students in the condition of active highlighting was also measured. Results: From linear mixed-effects models, the main effect of condition did not emerge for any of the outcomes. However, an interactive effect of condition and cognitive reflection emerged for literal text comprehension that favoured readers in the condition of experimenter-provided highlighting with higher ability to resist automatic thinking. Inferential text comprehension, transfer of knowledge, and calibration of performance were only predicted by cognitive reflection or reading self-efficacy. Finally, the quality of information highlighted significantly contributed to students' literal text comprehension and transfer of knowledge in the learner-generated highlighting condition. Takeaways: Active highlighting is not effective per se during digital reading. The "amplification" effect of already highlighted text and higher cognitive reflection suggests that readers who are more able to resist automatic thinking may also invest more effort in the task, taking more advantage of the provided support. Even if active highlighting may not be effective per se compared to other reading conditions, what students highlight contributes to literal text comprehension and their learning from text.