Industrialization experiences differ substantially across countries. We use a benchmark model of structural change to shed light on the sources of this heterogeneity and, in particular, the phenomenon of premature deindustrialization. Our analysis leads to three key findings. First, benchmark models of structural change robustly generate hump-shaped patterns for the evolution of the industrial sector. Second, heterogeneous patterns of catch-up in sectoral productivities across countries can generate variation in industrialization experiences similar to those found in the data, including premature deindustrialization. Third, differences in the rate of agricultural productivity growth across economies can account for the majority of the variation in peak industrial employment shares.