The existence of a genetic background is well admitted in schizophrenia, but some individuals at genetic risk for that disease could never manifest it at a clinical level. However, several vulnerability models could help us to identify such individuals. According to them, when similar perturbations at a given task are observed both in clinically stable patients with schizophrenia and their nonschizophrenic first degree relatives, this task could be qualify as an indicator of the vulnerability to schizophrenia. In literature, that seems the case for auditory ERP late components in oddball paradigms. Our study was undertaken to replicate literature data. For that purpose, amplitude and latencies of auditory N100, P200, N200 and P300 wave-forms were assessed among 21 clinically, stable schizophrenics, 21 of their biological full siblings and 21 unrelated control subjects matched with the two others groups for several socio-demographic factors. Comparison were performed by non parametric analyses (Kruskal-Wallis one way ANOVA, and post-hoc Mann-Whitney). Compared to controls, delayed latencies and/or reduced amplitudes were observed for several ERP components--mainly with P300--in the sibling group. ERP values from this group did not statistically differ from those of the group with schizophrenia. In conclusion, results from the sibling group suggest that ERP impairments in auditory oddball paradigms may actually represent indicators of the genetic vulnerability to schizophrenia.