There are 2 principal techniques of functional MRI (fMRI): the blood-oxygen-level dependent (BOLD) technique, which is the favoured method because no intravenous contrast medium is required, and the dynamic or exogenous technique. The BOLD technique takes advantage of the fact that the change from diamagnetic oxyhemoglobin to paramagnetic deoxyhemoglobin that takes place with brain activation results in decreased signal intensity on MRI. Commercially available scanners can be used to conduct single-slice BOLD fMRI experiments, but echo-planar hardware is needed for multislice wholebrain experiments. Sequence choices in BOLD fMRI include spin-echo and gradient-echo sequences, to which rapid acquisition with relaxation enhancement and echoplanar techniques may be applied. Optimal imaging parameters (echo time, slice thickness, field of view and flip angle) are important in maximizing signal-to-noise ratios. Various statistical techniques and software programs have been developed to interpret the large amounts of data gathered from BOLD fMRI experiments, which presents one of the biggest challenges in performing this technique with clinical MR units. Controversy exists regarding the effects of draining veins on cortical mapping, of inflow of blood into the imaging slice or volume, and of motion artifact. BOLD fMRI has demonstrated good correlation with positron emission tomography, magneto-encephalography and electrocorticographic recordings of motor responses. It has been used to study cortical activity of visual, motor, auditory and speech tasks as well as brain centres for smell, motor imagery, complex motion and memory. As such, it holds promise for the study of brain function, but must be subjected to larger studies comparing it with the gold standard of electrocorticographic mapping.