This paper presents the qualitative results of a study of students' reading of multimodal texts in an interactive, online environment. The study forms part of a larger project which addressed image-language interaction as an important dimension of language pedagogy and assessment for students growing up in a multimedia digital age. Thirty-two Year 6 students representing a sample of high, medium and low performers on an Australian state-wide school literacy test were surveyed about their internet usage and interviewed using a structured protocol while working online through a selection of materials from an educational website. Findings from the earlier stages of the project indicated that different types of image-text relations vary in the degree of difficulty they pose for students' reading comprehension. This phase of the project extended the analysis of image-text relations to online, interactive texts. Student performance on online reading tasks and interview data are used to illustrate some of the complexities students encounter when reading online, and how this may vary with factors such as their day-to-day literacy experiences and levels of engagement. The results have implications for literacy pedagogy and assessing the reading of web-based texts.