The prevalence of top-heavy bureaucracies in non-democracies cannot be explained by the theories of Parkinson, Tullock, Niskanen, or Simon or by classical managerial theories. When bureaucracy positions carry rents, the competition for promotion becomes a rent-seeking process. Borrowing the career-tournament theory framework from managerial scholarship, we argue that top-heavy bureaucracy resembles a tournament with too many finalists. When rent is centralized at the top (i.e. power centralization), as is the case in many non-democracies, the optimal bureaucracy should be top-heavy, accommodating and encouraging relatively more finalists at the top to compete for the final big prize. We provide suggestive evidence by analyzing ministry organizations in China (1993–2014) and Russia (2002–2015). After some fluctuations, the shape of Russian ministries eventually converged with that of China. In the steady state, their ministry shapes are far more top-heavy than what is prescribed by managerial theories. At the micro-level, ministry power centralization, measured by the perceived influence of the ministers, is correlated with ministry top-heaviness in Russia.