The question as to whether a failure of recognition unrelated to impaired sensory processing or to disorder of naming can occur in the somato-sensory modality has been eagerly debated in the french neurology. Taking as an argument the fact that he had never observed a tactile agnosia in the absence of subtle sensory deficits Dejerine denied the localizing value of tactile agnosia (or asterognosis). Conversely Delay, 20 years later, identified tactile performances such as discrimination of texture and shapes, which he considered as a specific neocortical function, that were lost in parietal syndromes with astereognosis and preserved elementary sensations. He also coined the term "tactile asymbolia" to qualify the patients with astereognosis in whom these performances are preserved. When referring to the definition of agnosias only "tactile asymbolia" should be considered as a "true" tactile agnosia. The recording of early somatosensory evoked potentials (SEPs) now offers the possibility of assessing non invasively the function of the primary somatosensory cortex (in particular area 3b). We have recorded SEPs to median nerve or finger stimulation in 309 subjects with a focal hemispheric lesion presenting with a somatosensory deficit of any type. We could confirm that asterognosis referable to impaired discrimination of textures and/or shapes in the absence of impaired elementary sensation is quite rare since it was observed in only 12 of our patients (3.9%). Moreover early cortical SEPs reflecting the activity of the primary somatosensory area (N20 or/and P27) were clearly abnormal in all of them. A single patient of this group of 12 could be considered as a case of tactile asymbolia but his early cortical SEPs were abnormal. The only condition combining a failure of tactile recognition of objects with normal early SEPs is represented by the "tactile anomia" observed in callosal dysconnexions. Thus, in our patients unable to identify objects by palpation in spite of preserved elementary sensation, we were unable to identify a case of pure tactile agnosia. These results are to be confronted with the recent observation by Caselli (1991, 1993) that tactile object recognition is partially, but selectively, impaired in inferior parietal lesions supposed to disconnect the ventrolateral associative somatosensory cortex from its limbic targets.