Purpose: The aim of the study was to investigate the effectiveness of social communication skills training (TBIconneCT) for people with traumatic brain injury (TBI) and their communication partners, delivered in-person or via telehealth, on quality of conversations. Method: This study is a clinical trial, including an in-person intervention group (n = 17), a telehealth intervention group (n = 19), and a historical control group (n = 15). Participants were adults at least 6 months post moderate-to-severe TBI with social communication skills deficits and their usual communication partners. Participants completed a casual and purposeful conversation task at pre-intervention, postintervention, and a follow-up assessment. A blinded assessor evaluated conversations using the Adapted Measure of Participation in Conversation and the Adapted Measure of Support in Conversation. Treatment effects were examined by comparing groups on change in ratings between pre- and posttraining. Maintenance of effects was examined using change between posttraining and follow-up assessment. The trial protocol was registered with the Australian New Zealand Clinical Trials Registry (ACTRN12615001024538). Results: Trained participants with TBI had significant improvements in participation in casual conversation compared to controls. Trained communication partners also had significant improvements compared to controls on ratings of support in casual conversations. However, treatment effects were not maintained at follow-up for two of eight measures. Comparisons between outcomes of in-person and telehealth groups found negligible to small effect sizes for six of eight measures. Conclusions: The findings reinforce previous studies demonstrating the efficacy of communication partner training after TBI. Telehealth delivery produced similar outcomes to in-person delivery.