In this paper, we reviewed a new data-driven analysis method, which can be used in the naturalistic paradigms: Inter-Subject correlation (ISC). ISC is a simple approach to analysis functional Magnetic Resonance Imaging (fMRI) data collected under naturalistic stimuli because this method doesn’t require a hypothesis about time course based on the task. Most neuroimaging studies like fMRI are trying to investigate the complex functions of human brain using the simple static stimuli. While in this highly controlled and simplistic experimental situation, the human brain may not process information just as exactly in a natural situation. Because of lacking ecological validity in traditional neuroimaging studies, there is an increasing trend in studying the human brain function under dynamic, continuous stimuli in order to be closer to normal everyday life than in conventional, controlled paradigms. But this kind of data can’t be analyzed by a hypothesis-driven analysis method like general linear model (GLM), because of the reason that GLM requires a reference time course of task (hemodynamic response function, HRF), this reference time course is defined as the expected time course of the BOLD signal (blood oxygenation level dependent) after the onset of a stimulus. Thus, approaches providing a more general flexibility are needed as alternative and complementary tools in analyzing fMRI data collected in a more natural situation or task. The basic hypothesis of ISC is that there would be a highly inter-subject synchronization cortical activity in related brain areas when individuals all are exposed in the same situation or doing the same task. By calculating correlation coefficient of time course based on the voxels in the same space location across subjects, the extent of shared processing across subjects can be estimated. Combining with other technics like reverse-correlation or others, ISC could be an appropriate way to investigate the relation between structures and functions of human brain. The validity of this non-parametric method has been proven to be suitable in natural situation against stimulus-model based analysis. The difference between ISC and conventional fMRI data analysis methods is that the former circumvents the need to specify a prior model for the neural processed in an given brain area, instead of which ISC compares the temporal response patterns across subjects in response to the identical stimulation (e.g. free viewing of movie or listening of conversation). Because of it's free-natural paradigm, which makes it more convenient to expand fMRI study into children or special groups, even non-human subjects. This novel exploratory method for fMRI data can be combined with other signal statistic analysis method (e.g. discrete wavelet transform) in order to investigate the data collected more deeply. In our daily life, cognitive processes (e.g. reading a book, engaging in conversation) unfold mostly over relatively long time scales. ISC makes it easier to investigate these large-scale cognitive processes like in real life. This review provides the theory and method, validity and the advantages of ISC in the research of brain imaging. By introducing the latest and significant findings of ISC’s application in neurobiology of cognitive, ISC was proposed as the growing trend toward cognitive neuroscience for more realistic and natural studies.