Due to a rapidly aging society, stagnating per capita income and rather stable prices, traditional economic analyses methods lost their power in the 1990s to predict future food product consumption in Japan. This study, in part, remedies this problem by projecting future consumption of selected products, using a cohort approach with economic factors tentatively set aside. Per capita consumption of individual household members by age was derived by incorporating family age composition into household data classified by age groups of household head (HH). Individual consumption estimates were decomposed into age, cohort and time effects, using the Nakamura's Bayesian cohort model. These effects were synthesized to predict per capita consumption in each population age cell in 2010 and 2020. Although some method refinement is warranted, the cohort approach proves a useful tool in improving prediction of future food product consumption and, combined with economic factors, may prove useful in future economic forecasting.