Over their life course, people change their consumption habits when prices, income, tastes or nutritional needs change. The time period during which an individual grew up is often reflected in his or her consumption of different types of food. To investigate the possible links between demographic changes and food consumption, we constructed two-step censored demand systems for different groups of foods. We estimated the systems using Norwegian data for the 1986 – 2012 period. In the systems, age, period, cohort, other demographic and economic variables are included. The estimated systems are used to construct a long-run forecasting model for meat and dairy products. In this model, younger cohorts replace older cohorts with a different consumption pattern. The total purchases of beef, lamb, pork and fluid milk are predicted to decrease, while the total purchases of chicken, yoghurt and cheese are predicted to increase towards 2027.