The purpose of this study is to investigate whether demand for foods in urban China is heterogeneous. Under the assumption of the two-stage budgeting, 18 food-at-home items are divided into five sub-groups. As a result, we estimate six separate quadratic almost ideal demand systems for the two stages, which incorporate demographic variables using the ”ordinary budget share scaling and translation” (OBSSAT) and a two-step estimator to deal with zero consumption problems when necessary. Employing Chinese urban household data in three coastal provinces in 1998, our empirical results show that, first, demographic variables are as important as economic factors in explaining food demands using the OBSSAT. Regional heterogeneity is strongly supported by using household data from three coastal provinces; therefore, China should be treated as several markets instead of one huge market. Second, income is still one of the most important economic factors affecting food demands. As household income consistently increases in urban China, our findings imply that the poor and the rich would react differently about their food consumption in relation to an increase in income. Finally, 18 at-home food items are investigated in this study; these food items also respond heterogeneously to prices, income, and demographic variables.