Purpose – The purpose of this paper is to construct a weight model from the sales comparison approach. Design/methodology/approach – Although weighted average of comparables into sales comparison value is commonly applied in the past, most papers only focus mathematical calculation. This paper examines the correlation between weight and attributes of 6,345 sales comparable properties adopting the multiple regression model. Findings – This paper finds the price type, proximity of transaction date, inside the neighborhood area or not, total gross adjustment as percent, numbers of adjustments and the attributes of other comparables considered in one appraisal are significant on the weight of comparables. The expected MAPE and Hit rate criterions are passed after forecasting 10 percent validation samples modeled by 90 percent samples randomly surveyed. Practical implications – The weighted average to determine the sales comparison value is reasonable since the value conclusion will “correlate” to indication of value derived by different comparables. Originality/value – This paper discusses the weight model and forecasts weights directly instead of only forecasting value. By elaborating on the core question of weights, this paper hopes to assist the degree of science and objectivity of appraisal.