We conducted a retrospective, descriptive evaluation of the fidelity with which parents of three children with autism spectrum disorders conducted functional communication training (FCT) in their homes. All training was provided to the parents via telehealth by a behavior consultant in a tertiary-level hospital setting. FCT trials coached by the behavior consultant were conducted during weekly 1-h visits. Parents made video recordings of treatment trials in which they conducted the procedures independent of coaching. We evaluated the levels of fidelity during coached and independent trials within a multielement design and recorded parents' omission and commission errors during different components of the treatment over time. The results showed no consistent differentiation between the coached and the independent trials. Some errors (e.g., omission errors associated with reinforcing manding) occurred more frequently overall, but none of the errors appeared to have a strong relationship with treatment outcomes. All children showed substantial reductions in problem behavior during the final treatment trials and especially during the coached trials. These results suggest that behavior analysts can use telehealth to implement FCT with acceptable fidelity and to achieve substantial reductions in children's problem behavior.