This study examined the effects of providing reading strategy instruction to improve the effectiveness of self-explanation (i.e., explaining the meaning of information to oneself while reading). The effects of the reading strategy instruction, called Self-Explanation Reading Training (SERT), were examined both in terms of comprehension scores and self-explanation quality. Half of the participants (n = 42) received SERT, which included reading strategy instruction and self-explanation practice with 4 science texts (SERT condition). The remaining participants read aloud the 4 science texts (control condition). During this training phase, self-explanation, as compared to reading aloud, only improved comprehension for the most difficult of the 4 texts. Prior domain knowledge consistently improved comprehension performance, whereas reading skill and reading span had minimal effects. After training, both SERT and control participants self-explained a difficult text about cell mitosis. SERT improved comprehension and self-explanation quality only for participants with low domain knowledge. However, the effects of SERT on low-knowledge participants' comprehension emerged only for text-based questions and not for bridging-inference questions. Protocol analyses indicated that SERT helped these participants to use logic, or domain-general knowledge, rather than domain-specific knowledge to make sense of the text.