Only a few studies have investigated neurophysiological substrates of phonological impairment in patients with aphasia (PWA) in the acute stage after stroke (Ilvonen et al., 2003; Nolfe et al., 2006). Behavioral evaluation is often problematic or even impossible in PWA in the acute stage, as some patients cannot be instructed due to severely impaired comprehension, reduced consciousness or confusion. Event-related potentials (ERPS) can circumvent such problems as they have already demonstrated their sensitivity and usefulness in measuring certain language processes in both a healthy and clinical population. The objective of the present study is to investigate neurophysiological substrates of phoneme and word processing in PWA with phonological disorders (PWA-PD) in the acute stage after stroke. Ten PWA-PD (5 men, 5 women; mean age 69.4 years +/- 3.46) are included and compared to 44 healthy control participants (HC) (20 men, 24 women; mean age 44.46 years +/- 13.76). All patients suffer from a first-ever stroke in the left hemisphere, are right-handed, have Dutch as native language and present with acute phonological disorders as established with the Psycholinguistic Assessment of Language Processing in Aphasia (Bastiaanse et al., 1995). PWA-PD admitted with a recurrent stroke, left handedness, indications for comorbid cognitive disorders and severe hearing deficits are excluded. Phoneme discrimination is studied in a pre-attentive (MMN) and attentive (P300) oddball task with respect to the phonemic contrasts place of articulation (PoA), voicing and manner of articulation (MoA) to explore whether a qualitative pattern of impaired phonemic contrast sensitivity can be determined. Word recognition is studied in a pre-attentive oddball task, which consists of differentiating real words from pseudowords. The electroencephalogram (EEG) is recorded through 23 Ag/AgCl-electrodes using a linked ears reference and an electrode placed on the forehead as ground. Further EEG analysis includes additional filtering, independent component analysis, segmentation, baseline correction and artifact rejection. Statistical analysis is performed on amplitudes and latencies specifically taking into account the large heterogeneity among PWA-PD. During phoneme discrimination, PWA-PD only show MMN amplitude reductions with voicing as phonemic contrast in the pre-attentive condition, whereas all three phonemic contrasts reveal smaller P300 amplitudes compared to HC in the attentive condition. PWA-PD show a larger response to PoA compared to MoA and voicing in the pre-attentive condition, whereas in the attentive condition only the difference between PoA and voicing remains. During word recognition, PWA-PD and HC display larger responses to pseudowords compared to real words from 100 ms onwards, continuing in the P200 and N400 time windows, despite the fact that responses to pseudowords show longer latencies in PWA-PD. In summary, this demonstrates a distinct pattern of impaired phonemic contrast sensitivity in PWA-PD, with PoA being the most resistant, voicing the most vulnerable and a substantial effect of attention. Moreover, PWA-PD suffer from a delay in lexical access due to a less efficient information transfer, which did not impair the response to pseudowords. For possible clinical implementation of ERPs, pre-attentive tasks seem to be more suitable than attentive tasks in the acute stage of aphasia.