The aim of this study is to evaluate whether SEP spatial mapping can improve outcome prediction in comparison to the conventional SEP recordings. Twenty patients comatose as a result of head injury or cerebral vascular disorders were submitted to 19-channel SEP mapping from median nerve stimulation. SEP recording were performed within the 4th hospital day in 18 cases and over one month from the insult in the remaining two. Nine patients (45%) showed a good recovery or a mild disability, 3 (15%) a severe disability and the rest (40%) died or remained in a vegetative state. Five patients (28%) had bilaterally normal SEP, 5 (28%) the absence of both parietal N20 and frontal N30, while the others (44%) had a dissociation N20/N30 (namely, preserved N20 with absent N30). The SEP mapping was significantly related to the outcome (P = 0.0087) and improved the outcome prediction in comparison to the conventional SEP recordings, allowing to check the presence of frontal N30: in patients with bilaterally present N20 the outcome appeared to depend upon the N30. SEP mapping proved to be a far superior prognostic indicator than the Glasgow Coma Scale. In 3 patients with midline shift on CT scan an abnormal spatial distribution of N20 was disclosed by SEP mapping. Our preliminary results suggest that SEP mapping may improve the assessment of comatose patients in comparison to the use of parietal derivations only.