Information theory is a probabilistic framework that allows the quantification of statistical non-independence between signals of interest. In contrast to other methods used for this purpose, it is model free, i.e., it makes no assumption about the functional form of the statistical dependence or the underlying probability distributions. It thus has the potential to unveil important signal characteristics overlooked by classical data analysis techniques. In this review, we discuss how information theoretic concepts have been applied to the analysis of functional brain imaging data such as functional magnetic resonance imaging and magneto/electroencephalography. We review studies from a number of imaging domains, including the investigation of the brain's functional specialization and integration, neurovascular coupling and multimodal imaging. We demonstrate how information theoretical concepts can be used to answer neurobiological questions and discuss their limitations as well as possible future developments of the framework to advance our understanding of brain function.