Functional magnetic resonance imaging (fMRI) studies in humans have recently provided evidence of intrinsic activity that propagates throughout the brain with non-zero time lags. Nevertheless, the limited temporal resolution of fMRI does not allow to characterize this lagged propagation on time scales relevant to behavior, i.e., in the 10-100 millisecond range. Thus, a millisecond resolution technique such as magnetoencephalography (MEG) is needed. Here we aim at investigating the spatiotemporal structure of intrinsic propagated activity on 61 resting state MEG subjects taken from the Human Connectome Project (HCP) database. For this purpose, we relied on a multivariate connectivity approach based on the cross-correlation between the amplitude envelopes of the alpha frequency band signals associated with cortical areas. Our findings clearly revealed that patterns of time lags between regions exist on temporal scales of tens of milliseconds, thus supporting the notion that communication by relaying signals is a central mechanism implemented by the brain. Finally, we discriminated between cortical regions that initiate the intrinsic activity propagation (sources) and cortical regions that act as destinations of the activity propagation (sinks). We found significant sources in the bilateral parietal areas and in the precuneus, and significant sinks in the left inferior temporal cortex.