The purpose of the present study was to assess the impact of coaching with performance feedback from teachers on accuracy of paraeducators' momentary time sampling (MTS) data of students' on-task behavior. Two lead teachers and three paraeducators participated in the study. The relation between coaching and accuracy of the data collection was evaluated using a multiple-baseline across paraeducators design. Baseline data from this study suggest that some paraeducators need explicit instruction in how to collect data with fidelity. Once coaching with performance feedback from teachers was implemented, there was an immediate increase in accuracy of data collection by paraeducators, as measured by inter-rater agreement, indicating a functional relationship between the independent and dependent variables. These results highlight that teacher-led coaching is feasible and effective for increasing paraeducators' MTS data collection accuracy and provides preliminary evidence that the accuracy maintained 1-2 months following cessation of formal coaching sessions. Implications for practice and future research are discussed.