Background: The refinement of nursing diagnosis has been recommended by the North American Nursing Diagnosis Association along years. Validation studies can play an important role in the process of refinement once it can give additional information whether the diagnosis can be considered adequate to detect a given phenomenon with validity and precision or if it still needs improvements. The Impaired Memory Nursing Diagnosis (IMND) has been proposed in 1994 and up to the moment no study assessed the validity of the IMND when determined by informant interview. The validation of this diagnosis to detect impaired memory of a subject by collateral source is important once the IMND is frequent in patients with cognitive decline, where data collection with caregivers are mandatory. Methods: From January to October, 2015, we interviewed 54 informants of autopsied patients, after informed consent, as part of the procedures of the Brazilian Brain Bank of the Aging Brain Study Group, in São Paulo – Brazil. For the investigation of clinically manifested cognitive decline in life we used the Clinical Dementia Rating Scale (CDR) applied to informants of the autopsied subjects, with reference to the ante-mortem period. CDR is a valid instrument and worldwide used to detect the level of cognitive decline in patients with dementia and was considered the gold-standard in the present study. Cognitive decline was considered present when CDR≥ 0.5. The existence of the IMND in each case was determined by Delphi Technique. The panel of specialists was composed by trained nurse specialists with experience in neuroscience and geriatric nursing. . Evidence of criterion validity was determined by the correlation between the IMND (present or absent) and Cognitive Decline (without CDR = 0; with CDR>0.5). For this, we used the Phi’s Correlation Test for nominal variables and we considered adequate evidence of criterion validity if the correlation was positive, significant (p<0.05) and > 0.50. Data was analyzed by SPSS v.21. Resultados: Almost half of the subjects had cognitive decline in life (40.7%). The IMND was present in 29.6% of the 54 cases and among them 93.8% had cognitive decline. IMND was present in 68.2% of patients with cognitive decline. On the contrary, 6.2% of the patients with IMND did not have chronic cognitive decline and 31.8% of patients with cognitive decline did not have IMND. Conclusão: In the present study, we could observe that IMND was frequently associated to clinically manifested chronic cognitive decline and the IMND presented strong evidence of criterion validity when assessed by informant interview. Nurses should pay special attention when determining the existence of this diagnosis in a given case once not all the patients with cognitive decline will have IMND and not all the patients with IMND will have it related to cognitive decline. Yet, the accuracy of the diagnosis will depend on the nurses’ ability to clinical reasoning and judgement.