Recent theoretical work on retail pricing dynamics suggests that retailers frequently change prices of specific items, even when their costs are unchanged. We extend this theory to explain which particular retail items will be subject to periodic temporary reductions. We then make use of a BLS data set on retail prices to document the frequency of sales and the specific products chosen for these temporary reductions across a wide range of goods and geographic areas. The results suggest that a number of pricing regularities exist for all 20 categories of goods we examine. First, retailers seem to have a "regular" price, and most deviations from that price are downward. Second, there is considerable heterogeneity in sale behavior across goods in a category (e.g. cereal); within each category of goods, the same items are regularly put on sale, while other items are rarely, if ever, put on sale. Third, the probability of a sale on an item appears to be greater when demand for that item is highest. Fourth, for the limited number of items for which we know category market shares, there is a statistically significant positive relationship between the likelihood a product is on sale, and its market share.