Teachers balance many goals when selecting texts for small-group reading. The author explored the impact of one variable that teachers consider, quantitative text difficulty, on students' participation, discussion, comprehension, and fluency in small-group discussions of informational science texts. Six bilingual third graders identified as reading below grade level participated. Students' comprehension scores were the same for matched and difficult texts. Matched texts benefited three students in terms of fluency and two students in terms of participation. Difficult texts benefited three students in developing conversations around print and ideas (rather than focusing on pictures) and in having more inferential (rather than merely literal) talk. Different levels of text had drawbacks and advantages, and the findings encourage teachers to use both matched and difficult texts at different times and for different instructional purposes.