Mass valuation is the valuation of a group of properties for the same purpose at the same time. It is applicable for property tax assessment however; in most developing countries, including Ethiopia property tax assessment is not clear and transparent. Thus, this study aims to customize the mass valuation model for the property tax assessment in Bahir Dar city. The study used a sample of 215 house rental value data which were recorded between 2011 to 2012, Ordinary Least Square was used to determine the physical and locational attributes that have a statistically significant influence on the house rental value. A stepwise regression method is used since it has the advantage to include the most significant variables in the model and makes the model simpler than complex. The information collected was analyzed using descriptive and inferential analysis. Through this study, using the significant variables, a mass appraisal model was developed for the study area. In the process, four models are developed and tested with statistical tests. The result reveals R2 statistics (.814) for neighborhood one indicates that the independent variables account for 81.4% of the rental value. Also, the R2 for neighborhood two .810 which implies that the independent variables account for 81.0% of the rental values. The R2 statistics for neighborhood three and the whole study area are (.813) and (.694) respectively. Meaning the independent variables account for 81.3% and 69.4% of the residential property rental value in neighborhoods three and four respectively and the statistical tests show this type of regression technique fits well for the residential property valuation. It was found that among the variables included in the model, number of rooms, availability of school, land value-grading, types of the nearest road, housing typology, built-up and plot area, walling and fencing materials appear as the most significant variables to predict the annual rental value of the residential properties in each neighborhood of the city. However, traveling cost and ceiling material of the house were not significant in influencing house rental value. The empirical pieces of evidence reveal that the individual models of each neighborhood are more accurate and provide more predictive power of the valuation model than the integrated model of the whole study area. Thus, the valuation models of each neighborhood can provide an effective valuation system. The validation of the model was evaluated using residuals, evaluation of adjusted R2, and Standard error of estimate, and it was found to be adequate. It was recommended that using this model in mass appraisal of residential properties in Bahir Dar in the future, potentially help to improve accuracy, objectivity, efficiency, and fairness of the property taxation system, which all these will lead to generating more revenue for the local government to finance their basic infrastructure and social services. Keywords: property valuation, property taxation, valuation model, mass appraisal, Ordinary Least Square