Heterotopagnosia-without-Autotopagnosia (HwA) is characterized by the incapacity to point to body parts on others, but not on one's own body. This has been classically interpreted as related to a self-other distinction, with impaired visual representations of other bodies seen in third person perspective (3PP), besides spared own body somatosensory representations in 1PP. However, HwA could be impacted by a deficit in the integration of visual and somatosensory information in space, that are spatially congruent in the case of one's own body, but not for others' body. Here, we test this hypothesis in a rare neurological patient with HwA, H+, as well as in a control patient with a comparable neuropsychological profile, but without HwA, and in age-matched healthy controls, in two experiments. First, we assessed body part recognition in a new task where somatosensory information from the participant's body and visual information from the target body shown in virtual reality was never aligned in space. Results show that, differently from the flawless performance in controls, H+ committed errors for not only the body of others in 3PP, but for all conditions where the information related to the real and the target body was not spatially congruent. Then, we tested whether the integration between these multisensory bodily cues in space, as during visuo-tactile stimulation in the full-body illusion, improves the patient's performance. Data show that after the stimulation prompting visuo-tactile integration, but not in control conditions, the patient's abilities to process body parts improved up to normal level, thus confirming and extending the first findings. Altogether, these results support a new interpretation of HwA as linked to the matching between somatosensory inputs from one's body and visual information from a body seen at a distance, and encourage the application of multisensory stimulation and virtual reality for the treatment of body-related disorders.