Organophosphorus neurotoxic agents (OPNAs) seriously damage the nervous system, inhibiting AChE activity and threatening human health and life. Timely and accurate detection of biomarkers in biomedical samples is an important means for identifying OPNA exposure, helping to recognize and clarify its characteristics and providing unambiguous forensic evidence for retrospective research. It is therefore necessary to summarize the varieties of biomarkers, recognize their various characteristics, and understand the principal research methods for these biomarkers in the retrospective detection of OPNA exposure. Common biomarkers include mainly intact agents, degradation products and protein adducts. Direct agent identification in basic experimental research was successfully applied to the detection of free OPNAs, however, this method is not applicable to actual biomedical samples because the high reactivity of OPNAs promotes rapid metabolism. Stepwise degradation products are important targets for retrospective research and are usually analyzed using a GC–MS, or an LC–MS system after derivatization. The smaller window of detection time requires that sampling be accomplished within 48 h, increasing the obstacles to determining OPNA exposure. For this reason, the focus of retrospective identification of OPNA exposure has shifted to protein adducts with a longer lifetime. Compared to the fluoride-induced reactivation method, which cannot be used for aged adducts, digestive peptide analysis is the more elegant method for detecting various adducts, identifying more active sites, exploring potential biomarkers and excavating characteristic ions. Retrospective identification of biomarkers after OPNA poisoning is of primary importance, providing unambiguous evidence for forensic analysis in actual cases and judgment of chemical accidents. At present, degradation products, the nonapeptide from BChE adducts and Y411 from human serum adducts are used successfully in actual cases of OPNA exposure. However, more potential biomarkers are still in the discovery stage, which may prove inconclusive. Therefore, there is an urgent need for research that screens biomarker candidates with high reactivity and good reliability from the potential candidates. In addition, mass spectrometry detection with high resolution and reactivity and an accurate data processing system in the scanning mode must also be further improved for the retrospective identification of unknown agents.