Improving educational outcomes involves many variables, including identifying effective interventions and ensuring that they are effectively implemented in schools. Within a "response to intervention" model, treatment integrity of academic interventions has become increasingly important. However, recent research has suggested that ensuring treatment integrity by instructional staff may require intensive coaching, including daily or weekly performance feedback. This system may be unsustainable in typical schools because of limited resources for supervision. Some studies have found that treatment integrity can be achieved with intense prior training that includes extensive practice followed by feedback in the training setting. Fluency-based instruction has the advantage of providing multiple practice opportunities in a relatively short amount of time. A fluency training package for paraprofessionals using the Corrective Reading: Decoding curriculum was evaluated in a multiple baseline design across individuals. The primary dependent variables included paraprofessionals' presentation rate and praise rate. Additional dependent variables included paraprofessionals' accuracy in presenting error correction procedures, ratio of positive to negative comments, students' on-task behavior, and word reading accuracy. Participants included five paraprofessionals delivering supplemental reading instruction to students in small groups, and one student from each of the paraprofessionals' groups. We provided five hours of fluency training to paraprofessionals over five days in a group setting. Following fluency training we observed paraprofessionals during a maintenance phase. Paraprofessionals generally increased their presentation rates, praise rates, and percentage of accurate error correction steps with fluency training. Three paraprofessionals with variable positive-to-negative comments ratios decreased this variability during fluency training. We subsequently provided performance feedback if a paraprofessionals' presentation rate or praise rate did not maintain at criterion levels. Four of the five paraprofessionals required performance feedback on at least one skill. Performance feedback had mixed effects on paraprofessionals' skills. Most students maintained adequate word reading accuracy throughout the study, with no clear effects when interventions for paraprofessionals were introduced and withdrawn. Students' on-task behavior was variable throughout the study, with decreases in variability for three students corresponding with fluency training for paraprofessionals.