This paper uses new data to reexamine trends in concentration in U.S. markets from 1994 to 2019. The paper's main contribution is to construct concentration measures that reflect narrowly defined consumption-based product markets, as would be defined in an antitrust setting, while accounting for cross-brand ownership, and to do so over a broad range of consumer goods and services. Our findings differ substantially from well established results using production data. We find that 42.2% of the industries in our sample are “highly concentrated” as defined by the U.S. Horizontal Merger Guidelines, which is much higher than previous results. Also in contrast with the previous literature, we find that product market concentration has been decreasing since 1994. This finding holds at the national level and also when product markets are defined locally in 29 state groups. We find increasing concentration once markets are aggregated to a broader sector level. We argue that these two diverging trends are best explained by a simple theoretical model based on Melitz and Ottaviano (2008), in which the costs of a firm supplying adjacent geographic or product markets falls over time, and efficient firms enter each others' home product markets. Institutional subscribers to the NBER working paper series, and residents of developing countries may download this paper without additional charge at www.nber.org.