Functional MRI (fMRI) has transformed academic neuroscience in the last decade and is now the most widely used non-invasive functional brain imaging technique. However, it has had much less of an impact in clinical neuroscience. While a majority of fMRI applications examine brain function in response to an externally driven task, an alternative approach characterizes the brain's intrinsic functional architecture. This involves fMRI scanning in the absence of an explicit task (i.e., in the resting state) and is, therefore, much more easily performed and tolerated by neurological and neuropsychiatric patient groups. The data are easily acquired, and the functional networks extracted are reproducible and reliable. However, quantifying networks of distributed brain activity and identifying the most informative features in a particular disorder remain a challenge. Progress has been made in this direction in recent years, with the adoption of mathematical tools from communications engineering. Specific alterations to the brain's functional connectivity at rest have been observed in generalized and focal epilepsies, as well as in non-epileptic attack disorder. The challenge for the future is to exploit knowledge of how the brain works as a complex system in order to develop more accurate and sensitive diagnostic tests for neurological and neuropsychiatric disorders.