An article residual value predicting device for accurately predicting the residual value of an article such as a used car. The device, for example, updates a set of basic records including the names of cars, used car prices, new car prices, and the years and months of the used car prices (S2), reads the used car prices and the new car prices, calculates the car residual value ratio actual values as their ratios (S6), performs regression analysis on the basis of the quantification theory type-I by using the car residual value ratio actual values as the response variables and the names of cars and the years and months of the used car prices as the explanatory variables, calculates the category scores (S9), and calculates the car residual value ratio predicted value to be predicted at a future time point as the scores-by-car name + scores-by-year + scores-by-month + constant from the category scores (S13). Prior to the regression analysis, the device carries out weighting processing (S4 to S8), copies the records according to the weights based on the then number of new cars sold, the number of car colors, and the number of records corresponding to them.