In this paper, our aim is to shed a new light on the analysis of REITs in the presence of time-varying exposures and errors-in-variables (EIV). From different multi-factor asset pricing models including the standard Fama-French-Carhart asset pricing model and the Pastor and Stambaugh model, we use the Kalman filter and show evidence of EIV in the dynamic factor loadings. Our approach revisits the dynamic return-based style analysis in the REITs industry in USA, UK and Canada. Our promising results clearly report significant improvement of factor loadings and an increase of the accuracy of the return risk sources for eight REITs styles.