The wide variation in measuring diversity of food consumption challenges on our study of how socioeconomic factors affect diet diversity. This essay is the first study in the economic literature that has examined if the estimated income and education effects on diet diversity are sensitive to how diet diversity is measured. Using data from the China Health and Nutrition Survey 2004-2009, this study constructs five diversity indexes to address two scenarios: i) indexes are constructed from the same indexing method but different food classifications, and ii) indexes are constructed from the same food classification but different indexing methods. The issue of endogeneity in estimating income effect on diet diversity is addressed by using instruments for household income. Regardless of which index is adopted, the estimates’ sign, statistical significance, and patterns over time and along the income distribution are robust. Income effect on diet diversity is significant, positive, but decreasing along the income distribution and over time. Low income people, particularly the poorest quintile, will receive the most benefit when their household income rises. Education, on the other hand, has small and even insignificant effects on the diversity of food consumption. However, indexes based on more disaggregate food groupings or derived from the counting method are found to be more income elastic. This study also argues that a count index using more detailed food classifications better reflects the health benefits from variety of food consumption in China since it captures the diversity both within- and across-food groups. It, thus, is a more appropriate measure of diet diversity for food and nutrition policies, as well as economic research that focuses on health aspect of dietary diversification.