Recently, the area of repetitive transcranial magnetic stimulation (rTMS) targeting neurological rehabilitation has been advanced as a potential treatment for post-stroke cognitive impairment (PSCI). However, the underlying mechanisms remains to be elusived. This study aims to figure out cerebral functional manipulation of rTMS in patients with PSCI through using the resting-state functional magnetic resonance imaging (rs-fMRI). Thirty patients with PSCI were recruited and randomly allocated into two groups: the rTMS intervention group and control group. The rTMS intervention group was given 20 min of 5 Hz rTMS (or control) over left dorsolateral prefrontal cortex (DLPFC) besides routine cognitive intervention training for 3 consecutive weeks, five times per week, on weekdays. Cognition performance was assessed by the Minimum Mental State Examination (MMSE) and Montreal cognitive assessment (MoCA). Neural activity and functional connectivity (FC) changes were acquired by rs-fMRI with fractional amplitude of low-frequency fluctuation (fALFF) and seed-based correlation analysis. Cognition improvements were observed both in rTMS intervention group and control group ( < 0.01), while the rTMS group got more significant improvent than control group ( < 0.05). To be specified, compared with the control group, the rTMS group got higher fALFF values in these brain regions including superior temporal gyrus, inferior frontal gyrus and parahippocampal gyrus, while lower fALFF values in middle temporal gyrus, middle frontal gyrus and fusiform gyrus. In addition, the rTMS group showed increased FC between LDPFC and toprecuneus, inferior temporal gyrus, middle and inferior frontal gyrus and marginal gyrus, while decreased FC between LDPFC and middle temporal gyrus and thalamus. The increase and decrease of neural activity and FC in cognition-related regions detected by rs-fMRI are good indicators to clarify the underlining mechanisms of rTMS on PSCI.