This paper examines the concept of "climate matching'' in migration—the idea that migrants seek out destinations with familiar climates—and studies its implications for the geography of economic activity in the United States. We document that temperature distance between origin and destination predicts the distribution of migrants across U.S. counties, for both internal and international migration in the historical (1850–1940) and modern (1970–2019) periods. These patterns cannot be explained by the spatial correlation of climate or the persistence of ethnic networks, and instead reflect two mechanisms: the transferability of climate-specific skills and climate as an amenity. We then study the economic consequences of climate mismatch during 1880–1920, a period of rapid growth and structural transformation. Using an instrumental variable strategy that interacts origin-country inflow shocks with the timing of county railroad access, we find that mismatch reduced agricultural productivity and accelerated the exit from farming. However, manufacturing output did not rise. Instead, manufacturing productivity declined and population growth was lower in counties with higher climate mismatch. These effects left a lasting imprint: a 1°C increase in 1880--1920 mismatch is associated with 2.5% lower per capita income in 1940.