Background: According to NANDA the ‘Impaired Memory Nursing Diagnosis’ (IMND) represents the “Inability to remember or recall bits of information or behavioral skills”. It is a frequent condition among elderlies. As it is mostly related to neurodegenerative diseases, its existence is determined by nurses after clinical judgment based on data often obtained by patient’s assessment and by informant interview, as well. Collecting data by collateral source may be difficult for nurses in a general basis, especially if the informant is facing moments of grief, like death of a close relative. It can be even more difficult if the assessment of the patient prior to death is lacking, as it is done in verbal autopsy procedures. As far as it is known, no other studies analyzed the level in which a nurse, during graduating phase, can correctly detect the existence of IMND by informants in moments of grief. Methods: From January-October 2015, four trained nurses (BNS; RN; master’s degree students in geriatric nursing) and one reference nurse (BNS; RN; CNS; Ph.D; professor of nursing) revised 54 clinical records of autopsied subjects, in order to detect the existence of the IMND. Comprehensive clinical data (health history; cognitive, behavioral and functional assessment) that composed the records were obtained by a knowledgeable informant right after the death of their next of kin (Postmortem interval of ±14h), after informed consent, with reference to the period prior to death. Cognitive assessment was specifically assessed by IQCODE and CDR. All the nurses searched the records for related factors and defining characteristics of the IMND, according to the current classification (2015-2017). The reference nurse detected the existence of the IMND in each case by “best estimated diagnosis”. The trained nurses analyzed each case in separate and emitted a clinical impression of whether the IMND was present or not, based on the existence of its related factors and defining characteristics. All the nurses were blinded to each other’s result. The diagnosis emitted by the reference nurse was considered the gold-standard and was used for comparison. For the analysis, we used Cohen’s Kappa to test the agreement between the reference nurse and each trained nurse, and we used the weighted Fleiss’s Kappa to test the overall interrater agreement between the four trained nurses. Agreement was considered good if Kappa > 0.61 (indicative of substantial agreement). Results: IMND was present in 29.6% of cases (n=16), as detected by the reference nurse. The trained nurses detected the IMND in 93.75% of times that it was detected by the reference nurse. When analyzing the concordance between the diagnosis done by the reference nurse and each trained nurse, it was observed an almost perfect agreement for each of them. Conclusion: Trained nurses can almost always correctly detect the existence of IMND by informants in moments of grief, with almost perfect agreement. These results may be particular interesting whenever considering the need to retrospectively determine cognitive decline by nurses.