PURPOSE – The purpose of this paper is to investigate dynamic food demand in urban China, with use of a complete dynamic demand system - DLES-LA/DAIDS, which pushes forward the techniques of demand analysis. DESIGN/METHODOLOGY/APPROACH – We employ a transitionary demand process and develop a new approach of complete demand system with a two-stage dynamic budgeting : an additively separable dynamic linear expenditure system (DLES) in the first stage and a linear approximate dynamic almost ideal demand system (LA/DAIDS). Employing provincial aggregate data (1995-2010) from the China urban household surveys (UHS), we estimated the demand elasticities for primary food products in urban China. FINDINGS – Our results indicate that most primary food products are necessities and price-inelastic for urban households in China. We also found that the dynamic model tends to yield relatively smaller expenditure elasticities in magnitude than the static models do due to dynamic adjusting costs, such as habit formation, switching costs, and learning process. PRACTICAL IMPLICATIONS – The research contributes to the demand analysis methodologically, and can be used for better projections in policy simulation models. ORIGINALITY/VALUE – This paper methodologically releases the restrictive assumption of instant adjustment in static models and allows consumers to make a dynamic decision in food consumption. Empirically, we introduce a new complete dynamic demand model and carry out a case study with the use of urban household data in China.