Soda taxes aim to reduce excessive sugar consumption. Their effectiveness depends on whether they target individuals for whom the harm of consumption is largest. We estimate demand and account for supply-side equilibrium pass-through. We exploit longitudinal data to estimate individual preferences, which allows exible heterogeneity that we relate to a wide array of individual characteristics. We show that soda taxes are effective at targeting young consumers but not individuals with high total dietary sugar; they impose the highest monetary cost on poorer individuals, but are unlikely to be strongly regressive if we account for averted future costs from over consumption.