One of the major challenges that a remanufacturer faces at strategic planning level today is to match its supply (returned items) with demand due to the inherited uncertainties and variations on both sides. Forecasting product returns is one of the most important tasks of this matching process. Unlike forecasting for traditional manufacturing systems, both quantity and quality forecasts are critical since return timing, quantity, and the quality of returned products can all vary dramatically. This research develops a forecasting method which incorporates knowledge from related sales, product usage, customer return behavior, and product life expectancy information to provide a more accurate prediction of product returns. The models are validated using Monte Carlo simulations. Numerical cases are also presented to illustrate its usage and some important insights.