This paper shows that consumers underreact to taxes that are not salient even in the long run, and reveals the trajectory of such effect over the number of store visits by consumers. Previous studies have suggested that tax salience effect may be persistent, but their evidence relied mostly on intertemporal variation in aggregate data. We use a novel consumer-level panel data in Japanese supermarket stores that include how the supermarket chains displayed their prices. The sample period coincides with the legal reform to relax the regulation that forced retailers to display the total price with sales tax. This regulatory reform provides a unique natural experiment, which allows us to estimate the causal effect on consumer behavior when sales tax is not shown. The difference-in-difference estimation shows that hiding a 5% sales tax increases individual sales value by 9.6% on a consumer's first visit after the change in price tags, but the treatment effect reduces sharply and stays stable around 3.8% on her second visit onwards. Consumers realize that price tags started to show only the base prices of products when they actually pay the total prices at cashiers on their first visits, which causes the sharp decline of the treatment effect. The fact that consumers' underreaction to sales tax remain after multiple store visits suggest that consumers intentionally ignore nonsalient taxes to avoid the cognitive cost of calculation. (JEL D12, H25, K34, L11)