There have been a number of empirical tests of the Linder hypothesis concerning how bilateral trade intensity is related to the per capita income similarity of two trading nations. However, due to the problem of multicollinearity which results from the high correlation between per capita income similarity and geographical proximity, it has been difficult to assess the significance of income similarity due to the overwhelming significance of geographical proximity. An alternative empirical test is proposed in this paper which allows the omitted variables and the correlated variables to be eliminated from the regression analysis. This test basically considers how the ratio of exports from two side-by-side nations with different incomes varies to a large set of importing nations equidistant from this pair of exporter nations. The model is formalized using the gravity model framework. The regressions use different sets of exporters including West Germany and Italy, Switzerland and Austria, Sweden and Greece, Japan and Korea, Sweden and Spain, and Australia and India. Different levels of commodity detail are also considered. Although overall support for the Linder hypothesis is weak, commodity characteristics generally associated with the Linder explanation such as the degree of product differentiation, human capital intensity, price responsiveness, income elasticity, and scale economies tend to be associated with Linder effects. One surprise is that the Linder effect is much more significant for capital goods than consumer goods; thus the modeling approach which focuses on consumer preferences appears misplaced since the emphasis should be on how a nation's income level affects the types of capital equipment desired.