Objective : To investigate the effects of cathodal transcranial direct current stimulation (tDCS) and continuous theta burst stimulation (cTBS) on neural network connectivity and motor recovery in individuals with subacute stroke. Design : Double-blinded, randomized, placebo-controlled study. Setting : University hospital rehabilitation unit. Participants : Inpatients with stroke (N=41; mean age, 65y; range, 28–85y; mean weeks poststroke, 5; range, 2–10) with resultant paresis in the upper extremity (mean Fugl-Meyer score, 14; range, 3–48). Interventions : Subjects with stroke were randomly assigned to neuronavigated cTBS (n=14), cathodal tDCS (n=14), or sham transcranial magnetic stimulation/sham tDCS (n=13) over the contralesional primary motor cortex (M1). Each subject completed 9 stimulation sessions over 3 weeks, combined with physical therapy. Main Outcome Measures : Brain function was assessed with directed and nondirected functional connectivity based on high-density electroencephalography before and after stimulation sessions. Primary clinical end point was the change in slope of the multifaceted motor score composed of the upper extremity Fugl-Meyer Assessment score, Box and Block test score, 9-Hole Peg Test score, and Jamar dynamometer results between the baseline period and the treatment time. Results : Neither stimulation treatment enhanced clinical motor gains. Cathodal tDCS and cTBS induced different neural effects. Only cTBS was able to reduce transcallosal influences from the contralesional to the ipsilesional M1 during rest. Conversely, tDCS enhanced perilesional beta-band oscillation coherence compared with cTBS and sham groups. Correlation analyses indicated that the modulation of interhemispheric driving and perilesional beta-band connectivity were not independent mediators for functional recovery across all patients. However, exploratory subgroup analyses suggest that the enhancement of perilesional beta-band connectivity through tDCS might have more robust clinical gains if started within the first 4 weeks after stroke. Conclusions : The inhibition of the contralesional M1 or the reduction of interhemispheric interactions was not clinically useful in the heterogeneous group of subjects with subacute stroke. An early modulation of perilesional oscillation coherence seems to be a more promising strategy for brain stimulation interventions.