A method for comparing estimated magnetoencephalographic (MEG) dipole localizations with regional cerebral blood flow (rCBF) activation areas is presented. This approach utilizes individual intermodal matching of MEG data, of rCBF measurements with [15O]-butanol and positron emission tomography (PET), and of anatomical information obtained from magnetic resonance (MR) images. The MEG data and the rCBF measurements were recorded in a healthy subject during right-sided simple voluntary movements of the foot, thumb, index finger, and mouth. High resolution 3D-FLASH MR images of the brain consisting of 128 contiguous sagittal slices of 1.17-mm thickness were used. MEG/MR integration was performed by superimposing the 3D head coordinate system constructed during the MEG measurement onto the MR image data using identical anatomical landmarks as references. PET/MR integration was achieved by a phantom-validated iterative front-to-back-projection algorithm resulting in one integrated MEG/PET/MR image. The estimated dipole locations followed the somatotopic organisation of the task-specific rCBF increases as evident from PET, although they did not match point-to-point. Our results demonstrate that intermodal matching of MEG, PET and MR data provides a tool for relating estimated neuromagnetic field locations to task-specific rCBF changes in individual subjects. Our method offers the perspective of refined dipole modelling.