The Norwegian purchases pattern of non-alcoholic beverages has changed substantially. The observed changes cannot be fully explained by changes in relative prices and income, and other variables are likely to be important. Cohort variables have not been given much attention in demand analysis. Our main contribution is to include age, period, and cohort (APC) variables into a demand system. The system is applied to purchases of non-alcoholic beverages. Wald tests show the importance of including APC variables. A Monte-Carlo analysis shows how replacing the APC variables with an age and a trend variable results in misleading age effects on milk purchases.