While most survivors have some spontaneous recovery after stroke, they reach a functional plateau and are left with persistent motor impairments. An emerging therapy, EEG-based Brain-Computer Interface (BCI), shows promise in promoting neural reorganization and facilitating additional motor recovery after stroke, however, the relationships between the neuroplastic changes and behavioral outcomes following this therapy are not fully understood. We examined changes in resting-state functional connectivity (rsFC) in the motor network and behavioral measures, including Stroke Impact Scale (SIS) and Action Research Arm Test (ARAT), over the course this therapy and investigated functional connectivity - behavior correlations. Twenty-six stroke patients with mild to severe upper extremity impairment completed EEG-based BCI therapy. Resting fMRI and anatomical scans were acquired on a GE 3T MRI Scanner before therapy, mid-, post- and one-month post-therapy, along with a neuropsychology battery. MRI scans of right hemisphere stroke patients were flipped to treat all subjects as left lesion hemisphere patients. We performed whole network, inter-hemispheric and intra-hemispheric seed region based connectivity analyses to study changes in rsFC over time and identify correlations between changes in rsFC and behavior over time. After therapy, a significant increase in network connectivity (p =0.000003) and intra-hemispheric connectivity (p = 0.047) from pre- to one-month post-therapy was observed. Additionally, inter-hemispheric connectivity increased and trended towards significance (p = 0.058). A significant positive correlation was observed between changes in network-level rsFC and SIS ADL (p = 0.005). Changes in intra-hemispheric rsFC correlated with ARAT for the affected arm (p = 0.001) and SIS Mobility (p = 0.003). The results suggest that EEG-based BCI therapy facilitates changes in rsFC in the motor network in stroke survivors, and these changes in connectivity are correlated with improvements in behavioral outcomes. This analysis provides a foundation for furthering our understanding of the potential of EEG-based BCI as a therapeutic modality for stroke rehabilitation to promote neurophysiological changes and motor recovery.