Identifying the objective biological markers is one of the most important scientific goals in psychiatric research. Recently, neuroelectrophysiological studies incorporating evidence-based medicine has provided an important approach and achieved abundant results to achieve this goal. However, there are still many controversies among current evidences, and it is difficult to reach a consensus. This article comprehensively summarized the results of psychiatric studies on mental disorders which combined the method of evidence-based medicine represented by meta-analysis and the neuroelectrophysiological technique represented by electroencephalography (EEG). To systematically analyze and summarize the neurocognitive dysfunction of mental disorders and high-risk groups, event-related potential (ERP) components are divided into early and late components including different cognitive activities such as perception processing, cognitive control, emotional response, and social cognition. Evidence based on quantitative EEG analysis are also discussed. We found that schizophrenia has defects in most of cognitive processes including perception, emotion and sociality, which indexed by amplitude reduction in most of the early and late components including P50, MMN, P300 and LPP. Attention deficit/hyperactivity disorder (ADHD) has defects in cognitive control indexed by amplitude reduction in the early and late components including the ERN and Pe. Anxiety and obsessive-compulsive disorder have defects in early cognitive control, while autism spectrum disorder (ASD) has defects in early perceptual processing and social cognition indexed by amplitude reduction of N170 and FN. In addition, abnormal P300 reflexing the impaired cognitive processing speed and attention resource allocation were found across multiple diagnosis types, which indicate that this ERP component may reflect the general cognitive deficits of mental disorders. The neuroelectrophysiological studies on evidence-based medicine for mental disorders has discovered a series of valuable ERPs or EEG objective marker features, but these features still cannot be directly used in the diagnosis of personalized medicine, until these features have been identified as objective biomarkers with high accuracy and specificity in further research and clinical practice. Future research can use multi-center database to seek the electrophysiological objective markers for mental disorders, and the integrated research combining advanced technologies including multi-modal psychoradiology and artificial intelligence methods including machine learning can further enhance the reliability and validity of biological markers of mental disorders.