A mixed-design study investigated how teachers and second/third- and fourth-grade bilingual students in three high-poverty schools responded to dialogic cognitive strategy instruction or dialogic responsive engagement (RE) instruction compared with a treated control (vocabulary). The second graders were taught in Spanish, and the fourth graders were taught in English. Qualitative analysis showed that it took time and additional support for the experimental teachers to move from whole-class instruction to small-group instruction. Although none of the second-grade teachers fully implemented the experimental treatments, the cognitive strategy second graders made significantly higher gains on a Spanish standardized reading test than those in RE. At the fourth-grade level, only one cognitive strategies teacher and one RE teacher fully implemented the experimental instruction. However, the cognitive strategy students performed significantly higher on a curriculum-embedded assessment and had a significantly higher gain on an English standardized reading test than students in the control group.