Curiosity is critical for learning and selective information processing, but it is unclear how it affects eye-movement patterns during reading. We tracked participants’ (n=56) eye movements in a task with 18 health-related questions. Each question was followed by up to 15 statements from three quality categories: scientific relevant (SR), scientific irrelevant (SI), and non-scientific relevant (NR). The statements were concealed in three boxes according to evidence quality. Participants could click on any box to reveal a new statement until they had read all the statements or were ready to answer the question. After reading a statement, participants rated their state curiosity and how useful the statement was for answering the question. The SR statements were rated highest on usefulness and were read most frequently, suggesting that participants were able to detect that these statements had the highest evidence quality. For a more comprehensive view on curiosity, we also measured participants’ trait curiosity with a questionnaire. We found that state curiosity correlated negatively with the thrill-seeking dimension of trait curiosity. Eye movements during statement reading were analyzed at the word-level, controlling for the effects of word length and frequency. Higher state curiosity was associated with lower word-skipping rates and longer total fixation duration on words, irrespective of evidence quality. However, the increase of total fixation duration with state curiosity was steeper for low quality (SI and NR) than high quality (SR) statements. Together, these results suggest that curiosity is related to a careful reading strategy, particularly for low-quality statements.