Nutrition transition is driven by quantity increase and structural change in food consumption. Particularly, meat consumption plays an important role. This study proposes a simple but innovative method to empirically decompose the total income effect on nutrition improvement into direct income effect and structural change effect, mediated by meat consumption share. With the use of the China Health and Nutrition Survey (CHNS) data, we find that a 1% increment in income will boost per capita calorie consumption by 0.02% within a family. The calories elasticity with respect to income is very small. However, 16 to 21% of the increase is due to dietary structural change, while the rest part is attributed to the conditional income effect. In addition, the dietary structural change effect is more prominent in the rural region, which implies a rural-urban gap in the diet.