This study examined whether reading a refutational or non-refutational text would induce different cognitive processing, as revealed by eye-movement analyses. Unlike a standard expository text, a refutational text acknowledges a reader's alternative conceptions about a topic, refutes them, and then introduces scientific conceptions as viable alternatives. Forty university students read one or the other type of text about the phenomenon of the tides. All had alternative conceptions about the topic. Findings showed that at post-test (off-line measure) refutational text readers learned more than non-refutational text readers. Outcomes regarding indices of visual behavior (on-line measures) during reading revealed that refutational text readers fixated the text segments presenting scientific concepts for a longer time overall than non-refutational text readers, in particular during the second-pass reading. Refutational text readers also fixated the refutational segments for a shorter time than non-refutational text readers for the control segments. Furthermore, all indices of visual attention predicted learning only for the refutational text readers. The more the students' reading of the refutational text was strategic, the better they learned from it. Implications about eye-tracking methodology and the refutational effect are drawn.