The effects on child engagement of adult use of systematic, naturalistic instructional strategies within an embedded instruction context were examined. Following baseline observations of two children with disabilities and same-age classmates without disabilities, three different instructional strategies were implemented with each child with a disability and one child without disabilities using an alternating treatments design. Results indicated that levels of engagement for the three target children during baseline (embedded instruction only) were low. Engagement was higher (and approximated levels of same-class peers without disabilities) for two of three systematic instructional conditions designed to increase the probability of responding by ensuring complete learning trials. Described are implications related to the need to use systematic, naturalistic instructional strategies in embedded instruction contexts and the potential use of engagement as an indicator of response opportunities.