This study examines children’s social responses to gender cues in synthesized speech in a computer-based instruction setting. Eighty 5th-grade elementary school children were randomly assigned to one of the conditions in a full-factorial 2 (participant gender) × 2 (voice gender) × 2 (content gender) experiment. Results show that children apply gender-based social rules to synthesized speech. More specifically, children evaluate synthesized speech more positively, trust the speech more, and learn more effectively when voice gender matches either content gender (consistency attraction) and/or their own gender (similarity attraction). Children’s computer self-efficacy was a significant covariate for their social responses to synthesized speech. Theoretical and practical implications of the current study for the design of educational media are discussed.