An analysis of the structure of demand was performed on household data classified into income groups for urban Indonesia. A demographically augmented Linearized Almost Ideal Demand System was used to estimate the structural parameters of the demand equations. Endogenous switching regression techniques yielded unbiased and consistent demand parameter estimates for the low income group, which had a large number of zeros for some food groups. Standard, seemingly unrelated, equation techniques were used to estimate the demand parameters for the other income groups. The results showed demands for the medium-high and high income households to be responsive to prices, income and demographic variables. Demands for the medium-low income households were responsive to income and prices only. Demands for low income households were responsive to income and prices of rice and fish only.