Previous studies suggest that macro- and micro-level factors jointly influence breastfeeding (BF) practices, but empirical evidence on the dynamics of such interactions along with the process of a country's economic development remains limited. Based on 42 Demographic Health Surveys (DHS) conducted in 15 Asian countries with a large time window (1990–2017), we thus test the existence of a reversal in the association between household wealth and BF practices throughout the development process. Four BF indicators (early initiation of BF, exclusive BF, continued BF at one year and two years) are examined, along with a standardized asset-based household wealth index allowing for cross-wave and cross-country comparisons. To highlight the dynamics of the wealth-BF association, we carry out econometric estimations, including interaction terms between household wealth and the country's level of economic development (low, medium, and high) or time. Instrumental variable estimations are also performed to limit suspected endogeneity issues. Our results confirm a transition in the wealth gradient of exclusive BF and continued BF in Asian countries. More precisely, while these practices are pro-poor in the poorest countries of the sample, they progressively spread to wealthier households along with the level of economic development. For exclusive BF, this transition has resulted in a reversal of the wealth gradient at the end of the period (i.e., exclusive BF prevalence among the rich overpassing that of the poor). We fail, however, to observe this kind of transition for early initiation of BF, this practice remaining pro-poor, whatever the level of economic development. To sum up, our results provide robust evidence of a transition in the wealth gradient of some BF practices along with economic development and time, and thus largely echo the literature exploring the social reversal hypothesis in the case of non-communicable diseases.< Réduire