The present study examined the neural markers measured in event-related potentials (ERPs) for immediate performance accuracy during a cognitive task with less conflict, i.e., a Stroop color-word matching task, in which participants were required to judge the congruency of two feature dimensions of a stimulus. In an effort to make ERP components more specific to distinct underlying neural substrates, recorded EEG signals were firstly dissolved into multiple independent components (ICs) using independent component analysis (ICA). Thereafter, individual ICs with prominent sensory- or cognitive-related ERP components were selected to separately reconstruct scalp EEG signals at representative channels, from which ERP waveforms were built, respectively. Statistical comparisons on amplitudes of stimulus-locked ERP components, i.e., prefrontal P2 and N2, parietal P3, bilateral occipital P1 and N1, revealed significant reduced P3 amplitude in error trials than in correct trials. In addition, significant evident ERN was also observed in error trials but not in correct trials. Considering the temporal locus of semantic conflict in the present task, we concluded that reduced P3 amplitude in error trials reflect impaired resolving process of semantic conflict, which further lead to a performance error in the Stroop color-word matching task.