In this paper is proposed a new method for functional connectivity detection using regression analysis. First, the resting state functional magnetic resonance imaging (fMRI) data from each voxel is passed through a bandpass filter to obtain frequencies between 0.01 Hz and 0.1 Hz. Then, the region of interest (ROI) is defined and the mean time course of all voxels in ROI is used as a regressor. Finally, the linear relationship between the time course of other voxels and the regressor in the resting brain is estimated. By the application of this new method in simulation data and fMRI data, the relevant validity and reliability are demonstrated.