Although the existing literature in economics and marketing offers growing evidence on the use of price points, there is a lack of direct evidence on the link between price points and price rigidity. We examine this issue in the retail setting using two datasets. One is a large weekly transaction price dataset, covering 29 product categories over an eight-year period from a large US supermarket chain. The other is from the Internet, and includes daily prices over a two-year period for hundreds of consumer electronic products with a wide range of prices. Across the two datasets, we find that 9 is the most frequently used price-ending for the penny, dime, dollar and the ten-dollar digits. Exploring the relationship between price points and price rigidity in these datasets, we find that the most common price changes are in multiples of dimes, dollars, and tendollar increments. When we econometrically estimate the probability of a price change, we find that 9-ending prices are at least 24 percent (and as much as 73 percent) less likely to change in comparison to prices ending with other digits. We also find that the average size of change of 9-ending prices are systematically larger when they do change, in comparison to non 9-ending prices. This link between price points and price rigidity is remarkably robust across the wide variety of price levels, product categories, and retailers examined in this study. To make sense of these findings, we offer a behavioral explanation building on the emerging literature on rational inattention. We argue that consumers may find it rational to be inattentive to the rightmost digits of retail prices because of the costs of processing price information. In response, firms may find it profitable to set the rightmost digits at 9, and therefore, be rigid in their pricing around these price points. We conclude that price points are a significant source of retail price rigidity in our datasets, and that rational inattention can offer a plausible explanation for their presence.