A common problem in food expenditure surveys is that many households report not purchasing some products. Usually the reason for these zero expenditures is unknown. The issue may be treated as a potential sample selection bias problem. This study estimates Engel functions for expenditures for fresh potatoes, processed products, and the value of home-grown potatoes with data from the 1989 Swedish Household Food Expenditure Survey. The two-step Heckman procedure is used. A strong generational pattern is found in potato consumption as well as several other significant effects.