Abstract There are current attempts to replace the WADA test for pre-surgical evaluation of hemispheric language capabilities by one of the methods of functional brain imaging. Recent PET and fMRI studies using verbal cognitive tasks like verb generation, semantic monitoring or semantic (`deep') encoding of words showed asymmetries of activation in the fronto-lateral cortex. In a previous ERP study subjects were required to indicate whether pronounceable non-words and abstract geometric figures were presented for the first time (`new item') or whether they had been shown before (`old item'). Group analyses of this study showed significant material-specific hemispheric asymmetries with ERPs being more negative-going in recordings of the posterior part of the left hemisphere with verbal material (CP5/6) but more negative-going in recordings of the right hemisphere with the spatial material (P7/8). The aim of the present study was to test statistically ERP lateralization effects in individual healthy subjects as well as WADA-tested patients suffering from seizures of the mesio-temporal lobe (MTL). In all subjects ERP lateralization with verbal material was tested in the electrode pair CP5/6, and ERP lateralization with figures in the electrode pair P7/8. Statistical analyses of single trials showed that in 20 out of 24 subjects ERPs with verbal material started to be more negative-going in CP5 as compared to CP6 in the period between 100 and 200 ms after stimulus onset or the subsequent time epoch (200–300 ms). In one subject not CP5/6 but the closely adjacent electrode pair P7/P8 showed this verbal material-related hemispheric effect. In patients language dominance as indicated by ERPs was not always consistent with the data of the WADA test. In one patient with left MTL seizures ERPs with verbal material and figures were found to be significantly lateralized to the right hemisphere although the WADA test assigned this patient to have a language-dominant left hemisphere.