Functional Magnetic Resonance Imaging (fMRI) scans provide a record of ongoing neural processing through the measurement of changes in regional levels of blood oxygenation. While subjects are resting, fMRI has revealed networks of brain regions exhibiting synchronised 0.02 - 0.12 Hz fluctuations in spontaneous, low frequency brain activity. The physiology underlying these 'resting state' fluctuations is poorly understood. Our current research investigates these networks within the human motor system, assessing their persistence over time, their covariance structure, and the effects that motor performance and training have on the networks. Here we present a review of this study, focusing on the methods involved in assessing the fMRI data, discussing issues of region identification, signal extraction, signal to noise, network covariance analysis and the detection of changes in the networks.