The purpose of this study was to test the reliability and validity of the Pain Assessment in Advanced Dementia (PAINAD) and particularly consider whether or not this measure was invariant when used among the Black and White residents. Baseline data from an implementation study testing that included a sample of 553 residents, 30% of who were Black, from 55 nursing were included in this study. The Winsteps statistical program was used to perform the Rasch analysis and evaluate the reliability and validity of the measure based on internal consistency, infit and outfit statistics, mapping, and a differential item functioning (DIF) analysis. The AMOS statistical program was used for confirmatory factor analysis. The findings supported the reliability and validity of the PAINAD when used with these individuals and demonstrated that there was no evidence of invariance between the Black and White residents. All the items fit the model, but there was not a good spread of the items across the pain level of the participants. The majority of the participants (75%) were so low in pain signs or symptoms that they could not be differentiated. Based on the clinical practice and observations, it is recommended that additional items can be added to the measure such as observing the individual for evidence of resisting care, retropulsion when trying to stand, hitting or kicking when turning in bed, hitting or kicking when transferring from bed to chair, hitting or kicking when ambulating, or hitting or kicking when raising arms, less engagement with others, and decreased participation in the activities previously enjoyed.