Objective To compare the effects of high-frequency (10 Hz) versus low-frequency (1 Hz) repetitive Transcranial Magnetic Stimulation (rTMS) on motor recovery and functional reorganization of the cortical motor network during the early phase of stroke. Methods Forty-six hospitalized, first-ever ischemic stroke patients in early stage (within two weeks) with upper limb motor deficits were recruited. They were randomly allocated to three groups with 10 Hz ipsilesional rTMS, 1 Hz contralesional rTMS, and sham rTMS of five daily session. All patients underwent motor function (Upper Extremity Fugl–Meyer), neurophysiological and resting-state functional Magnetic Resonance Imaging (fMRI) (rs-fMRI) assessments before and after rTMS intervention. Motor recovery (△Fugl–Meyer Assessment) was defined as motor function changes before and after rTMS intervention. Motor function assessment was reevaluated at time point of three month follow-up. Results The two real rTMS groups manifested greater motor improvements than the sham group. The effect sustained for at least 3 months after the end of the treatment sessions. Compared with the sham group, 10 Hz ipsilesional rTMS group presented increased resting-state functional connectivity (FC) between ipsilesional primary motor cortex (M1) and contralesional M1 (P = .007), whereas 1 Hz contralesional rTMS group presented increased FC between contralesional M1 and ipsilesional supplementary motor area (P = .010), which were positively correlated with motor recovery (P < .05). Conclusion Beneficial effect of rTMS on motor recovery might be underlaid by increased FC between stimulating site and the remote motor areas, highlighting the motor network reorganization mechanism of rTMS in early post-stroke phase.