This dissertation is focused on the role of income and the Linder effect in explaining the trade of 46 differentiated agri-food and beverage products across. 52 developed and developing countries from 1990 to 2000.The Heckman selection method and ordinary least squares (OLS) procedures were used to explore the role of income in explaining trade. The hypothesis that income does not influence agri-food and beverage product trade is consistently rejected. For most of the agri-food products, the empirical results reject the proposition that expenditure elasticities are the same across the development spectrum. The assumption of homothetic preferences (expenditure elasticities equal to one) was most consistently rejected for middle and high income countries when they trade with developed countries but rarely rejected for lower-income countries. However, in the case of middle-income countries, when preferences were nonhomothetic, the growth in imports of cereals, fresh and processed fruit and fresh and frozen fish outpaced their growth in income while this was never the case for high income countries. The results suggest that income plays an important role in agri-food trade and middle-income countries are the growth markets of the future. Using a generalized gravity equation, the study tests the Linder effect which states that as the demand structures of two countries become similar, their trade intensity increases. Two proxies, the Balassa (1986) index and the absolute value of the difference in per capita GDPs of trading partners were used to capture the Linder effect. Two measures of bilateral trade, the Grubel and Lloyed (GL) index, and the value of bilateral trade, were used. In addition, three forms of estimation were used to test the Linder effect: OLS with linear functional form including zero values of the GL index; OLS with log-linear functional form, ignoring zero values of the GL index; and Tobit estimation. The type of proxy used for the Linder effect and the way in which bilateral trade was measured influenced the outcome of the statistical tests of the Linder effect. However, the results do not provide convincing support for the effect.