A dominant tendency in cerebral studies has been the attempt to locate architecturally distinct parts of the cortex and assign special functions to each, through histological, clinical or hypothesis-based imaging experiments. Here we show that the cerebral cortex can also be subdivided into different components temporally, without any a priori hypotheses, based on the principle of functional independence. This states that distinct functional subdivisions have activity time courses (ATCs) that are, if not independent, at least characteristic to each when the brain is exposed to natural conditions. To approach a time-based anatomy experimentally, we recorded whole-brain activity using functional magnetic resonance imaging (fMRI) and analyzed the data with independent component analysis (ICA). Our results show that a multitude of cortical areas can be identified based purely on their characteristic ATCs during natural conditions. We demonstrate that a more "rich" stimulation (free viewing of a movie) leads to more areas being activated in a specific way than conventional stimuli, allowing for a more detailed dissection of the cortex into its subdivisions. We show that stimulus-driven functionally specialized areas can be identified by intersubject correlation even if their function is unknown. Chronoarchitectonic mapping thus opens the prospect of identifying previously unknown cortical subdivisions based on natural viewing conditions by exploiting the characteristic temporal "fingerprint" that is unique to each.