This paper provides evidence on trade barriers and their impact on trade volumes. It argues that progress in understanding trade barriers requires careful measurement, starting with direct data on freight and tariff rates. The paper presents three main contributions: (1) detailed data on freight rates for various importers, showing that freight costs are relatively low and that import choices are made to minimize transportation costs; (2) estimation of the technological relationship between freight rates and distance, revealing implausibly large trade barriers when interpreted through common proxies; and (3) use of a multi-sector model to isolate channels through which trade barriers affect trade volumes, allowing for direct estimation of substitution elasticities and a complete characterization of trade costs. The paper also provides stylized facts on freight and tariff rates, showing that freight rates vary significantly across exporters and commodities, with land-locked countries like Paraguay having exceptionally high freight rates. Freight rates are generally lower for manufactured goods than for commodities. The paper also compares freight and tariff rates, finding that freight rates are typically higher than tariff rates, with the exception of some manufactured goods in Latin American countries. The paper then discusses indirect methods for measuring trade costs, using a one-sector monopolistic competition model. It shows that trade barriers can be interpreted as a function of distance, language, adjacency, and border crossing. However, the paper argues that these proxies may not accurately capture trade barriers, as they may instead reflect preferences or production composition. The paper then estimates the freight-distance relationship, finding a distance elasticity of 0.27. It also discusses the implications of these estimates for interpreting proxy variables, showing that common proxies may imply implausibly large trade barriers. The paper concludes that these estimates suggest that explicitly measured costs (freight plus tariffs) are most of the story for many goods.This paper provides evidence on trade barriers and their impact on trade volumes. It argues that progress in understanding trade barriers requires careful measurement, starting with direct data on freight and tariff rates. The paper presents three main contributions: (1) detailed data on freight rates for various importers, showing that freight costs are relatively low and that import choices are made to minimize transportation costs; (2) estimation of the technological relationship between freight rates and distance, revealing implausibly large trade barriers when interpreted through common proxies; and (3) use of a multi-sector model to isolate channels through which trade barriers affect trade volumes, allowing for direct estimation of substitution elasticities and a complete characterization of trade costs. The paper also provides stylized facts on freight and tariff rates, showing that freight rates vary significantly across exporters and commodities, with land-locked countries like Paraguay having exceptionally high freight rates. Freight rates are generally lower for manufactured goods than for commodities. The paper also compares freight and tariff rates, finding that freight rates are typically higher than tariff rates, with the exception of some manufactured goods in Latin American countries. The paper then discusses indirect methods for measuring trade costs, using a one-sector monopolistic competition model. It shows that trade barriers can be interpreted as a function of distance, language, adjacency, and border crossing. However, the paper argues that these proxies may not accurately capture trade barriers, as they may instead reflect preferences or production composition. The paper then estimates the freight-distance relationship, finding a distance elasticity of 0.27. It also discusses the implications of these estimates for interpreting proxy variables, showing that common proxies may imply implausibly large trade barriers. The paper concludes that these estimates suggest that explicitly measured costs (freight plus tariffs) are most of the story for many goods.