Behavioral skills training (BST) delivered synchronously via telecommunication (telehealth) has been shown to be effective for training mediators supporting children with autism. A limitation in the literature is that many published training protocols are evaluated under analogue or highly controlled conditions; therefore, further evaluation is needed to assess generalization of training outcomes to other settings, with service recipients, and to untrained skills. Further, most studies have focused on reducing challenges vs. promoting skill acquisition. Using a concurrent multiple-baseline design, we assessed the efficacy of virtually delivered, synchronous BST with additional coaching (if mastery was not met) for training three behavior technicians (BTs) to implement two common behavioral protocols (i.e., errorless learning and paired stimulus preference assessment). Treatment integrity was assessed with the BTs implementing the protocols with an actor role playing a child with autism and with a child with autism receiving services in the same agency where the BTs were employed. Results indicated that the training was effective for increasing treatment integrity for all three BTs with their performance maintaining at follow-up; however, generalization outcomes were mixed. The findings have timely implications for conducting virtual behavioral training and add to the limited training research conducted in real-world settings.