This paper aims at simulating optimal prices satisfying public health recommendations in terms of nutrient adequacy. This implies to estimate a complete food demand system in order to compute price elasticities. Food consumption behaviors are described by an AI functional form [Deaton and Muellbauer(1980)] augmented to control for habit persistence. The demand system is estimated using the Iterated Least Square Estimator developed by Blundell and Robin (1999). We use French household expenditure data drawn from TNS Worldpanel covering 130 periods of 4 weeks from 1996 to 2005. Given the nature of our data, households are split into 8 cohorts based on two socio-demographic variables: date of birth and social status. A revised aggregation into 27 food groups is proposed in this paper. More precisely, commodities are grouped into homogeneous categories in terms of nutritional content and consumer preferences. Nutrient adequacy is defined using the MAR (Mean Adequacy Ratio), a nutrient only-based indicator. We calculate nutrient adequacy for 12 essential nutrients. Optimal prices are derived following Ramsey's approach to optimal taxation; Maximizing social welfare under nutritional constraints results in 27 optimal price variations or tax rates, each defined as a nonlinear function of all direct and cross price elasticities and the above mentioned indicator for all food groups.