Using a hedonic regression approach with data from 1.53 million Airbnb properties, I estimate the price of a representative short-term rental property at the center of 734 cities worldwide. The estimated rental prices provide an internationally standardized proxy for housing costs. Rental prices computed in this way are found to be highest in Amsterdam, London, New York, and San Francisco. I use these standardized rental price estimates to compute the elasticity of housing costs with respect to city size. My preferred specification shows an elasticity of 0.16, statistically significant at the 1% level. However, there is considerable geographic heterogeneity. Housing costs increase more strongly in city size in the euro area and India than elsewhere. In contrast, I find them to decrease in city size in Mexico. I offer suggestive evidence that crime might explain this unusual result.