Identifying individuals at risk of experiencing functional difficulty at home would support timely home safety assessment and modification services, which could lead to reducing home incidents such as falls. The objective of this study was to calculate older adults’ functional difficulty at home scores using the 12 physical function items in the American Housing Survey National and Metropolitan Data (AHS). Among the 28,474 older adults selected for this study, we used 19,932 for measurement model development and 8542 for model testing. Confirmatory factor analysis confirmed an adequate fit of the one-dimensional model with all AHS 12 items loading on one latent construct (functional difficulty at home) (RMSEA: 0.034, CFI: 0.990, and TLI: 0.988). Based on our model selection process, we determined that the Graded Response Model was an optimal model for our analysis and separated two detected differential functioning items for each sex. Using the testing dataset, we validated that the estimated functional difficulty scores showed an expected item hierarchy and statistically significant differences in their association with housing and demographic conditions (p < 0.001). Our results demonstrated the process of using the 12 AHS physical function at home items to produce validated scores of older adults’ functional difficulty at home.