There has been great emphasis, on how to reduce the consequences of food insecurity on the people of Ondo State, Nigeria, as a result of which there is a lot of research into demand for food. Estimation of demand for food has ignored required connection between theory and empirical analysis and concentrated on the estimation of single linear demand equation. Even where non linear models such as Almost Ideal Demand System (AIDS) models were used; there was no allowance for a non-monotonic relationship between the marginal budget share and total expenditure. Thus, this study examined demand for food in Ondo State using standard Quadratic Almost Ideal Demand System (QUAIDS) and the specific objectives are: (i) to examine the expenditure pattern, (ii) to determine how household demography affects household expenditure for food, and (iii) to analyse the difference in expenditure purchase among the households in the three senatorial districts of the State. Data collected from 1,200 heads of households, through multistage sampling methods were analyzed. Result shows that the QUAIDS test is more reliable, as the Wald test{Chi2(9)=340.71; Prob=Chi2=0.0000} indicates that lambda coefficients are jointly significantly different from zero and that the quadratic income terms are important, showing the superiority of QUAIDS model over the AIDS model. The estimated expenditure elasticities for all Ondo State are all positive and statistically significant at the 5%, indicating that all the food items are normal goods and that rice, beans, yam-flour, meat and vegetable and fruits are luxury goods since the coefficients are 1.419, 1.017, 1.385, 1.183 and 1.618 respectively which are greater than 1. However, garri, yam, bread and plantain are all necessity goods. The study conclude that policy-makers should consider consumer behavior at different income and price levels, as this will affect the rate at which people have access to food.
