Educational screen media is increasingly salient in the lives of young children. Research affirms preschool-aged children can learn content from media when they attend to it, however less is known about how specific screen-based pedagogical supports (SBPS) might draw children's attention. Using eye-tracking methodology, the current study examines specific SBPSs that engage children's attention. The sample consisted of 106 3- to 5-year-olds from a poverty-impacted neighborhood. Participants viewed 12 video clips of Sesame Street that used four different SBPSs to support vocabulary: visual effects, visual + sound effects, explicit definitions, and explicit definitions + repetitions. Results indicated that children attended significantly more to the SBPSs with definitions. Findings also revealed differences in screen composition. Children attended more to people than objects, and attended more to on-screen conversations than conversations cut between screens. This study demonstrates the importance for educational media to use appropriate SBPSs and on-screen compositions to engage children.