•Although workers in Germany often change jobs between industries in different sectors, inter-industry labor flows are highly structured.•Inter-industry labor flows span a sparse network: on average, industries representing 5% of national employment absorb 62% of labor-outflows.•This network, which can be seen as a reflection of industries’ skill relatedness, is similar for different worker types and barely changes over time.•Related employment defined by skill relatedness is a better predictor of local industry growth than by value chain or colocation based relatedness.•Because skill-related industries have uncorrelated growth rates, local employment shocks can be accommodated in skill-preserving ways. Using German social security data, we study inter-industry labor mobility to assess how industry-specific human capital is and to determine which industries have similar human capital requirements. We find that inter-industry labor flows are highly concentrated in just a handful of industry pairs. Consequently, labor flows connect industries in a sparse network. We interpret this network as an expression of industries similarities in human capital requirements, or skill relatedness. This skill-relatedness network is stable over time, similar for different types of workers and independent of whether workers switch jobs locally or over larger distances. Moreover, in an application to regional diversification and local industry growth, skill relatedness proves to be more predictive than colocation or value chain relations. To facilitate future research, we make detailed inter-industry relatedness matrices online available.