This review summarizes the main results obtained to date, concerning the issue of brain/language relationships, from studies using the modern techniques of in vivo functional imaging of the brain, namely Positron Emission Tomography (PET) and, to a lesser extent, the still experimental method of functional Magnetic Resonance Imaging. These studies mostly rely on the so-called cognitive activation paradigm, which consists of recording blood flow variations as measured in PET in normal volunteers during various linguistic tasks compared to reference tasks. Since the late 80's, a great deal of literature has been accumulating, including studies of various aspects of language functions: perceptual (auditory and visual) processing of linguistic materials, expression, verbal memory, semantics... The first few results obtained have already led to challenge some previously widely held dogmas, derived from classical anatomo-clinical studies, such as the role of Wernicke's area in verbal comprehension and the role of Broca's area in language production. The involvement of the left supramarginal gyrus in some aspects of phonological processing has been directly demonstrated. At the same time, some studies have yielded unexpected results suggesting, for instance, a special role for the inferior lateral parts of the left frontal lobe in semantic and/or voluntary aspects of language processing and episodic verbal memory. Among other studies, those concerning the mechanisms of verbal short-term memory have provided interesting parallels with cognitive models. The demonstration, under certain experimental conditions, of an involvement of hitherto unsuspected cortical regions (such as the insular or the temporopolar cortices) is of potential considerable importance. Finally, with current technical improvements of the imaging methods, it will be shortly possible to take into account the important issue of interindividual variations as well as to perform activation studies in brain-damaged patients.