We estimate the costs of occupational mobility using a novel approach that relies on aggregate flows of workers across occupations rather than on wage data. The theoretical underpinnings for this approach are derived from a model of occupation choice that delivers a gravity equation linking worker flows to occupation characteristics and to transition costs, which we proxy using task data from the Dictionary of Occupational Titles (DOT). Occupation flow data are constructed from the matched monthly Current Population Survey (CPS) between 1994 and 2012. We find that transition costs vary widely across occupations, are increasing in task distance (the dissimilarity in the mix of tasks performed in the two occupations) and are higher for transitions across broad task categories. However, most of the transition costs are accounted for by general, task-independent entry costs, specific to each destination occupation.