The authors investigated the effects of online discussion embedded within metacognitive guidance on mathematical literacy and self-regulated learning (SRL). They compared 4 instructional methods: online discussion embedded within metacognitive guidance (Online+Meta), online discussion without metacognitive guidance (Online), face-to-face discussion with metacognitive guidance (Ftf+Meta), and face-to-face discussion without metacognitive guidance (Ftf). Participants were 86 seventh-grade Israeli boys and girls who practiced problem solving in 4 classes. Results showed that the Online+Meta students significantly outperformed the Ftf+Meta students, who, in turn, significantly outperformed the Online and Ftf students on mathematical literacy of standard tasks, real-life tasks, and various aspects of self-regulated learning. Regarding Online students, results were mixed; these students outperformed the Ftf students on part of the criteria for standard problem-solving standard tasks and real-life tasks. The authors found no significant differences between the Online and Ftf students on SRL. Theoretical and practical implications of the findings are discussed. (Contains 7 tables.)