Hospital size distribution, i.e., the uneven allocation of healthcare resources between hospitals of different size classes, significantly impacts healthcare equality and efficiency and challenges healthcare governance. Yet, its variation is inadequately explained by two prevalent approaches to size distribution – the statistical approach merely works on statistical patterns, while the economic approach has difficulties in calculating hospital cost-returns. This study provides an alternative – a geographical approach – to demystify hospital size distribution. Incorporating the distinctiveness of healthcare into George Stigler's (1958) classic survivor analysis of firm size distribution, we conceptualize hospital size distribution as the result of hospitals' interactions with their socio-economic-technical environment in a defined area, and thus its pattern and variation could be understood through its coevolution with geographical contexts. We substantiate this idea with a study on China's provincial hospital size distributions between 2002 and 2019. During this period, hospitals with 0–99 beds maintained a stable proportion, while the size class of 800+ beds grew significantly, in contrast to the reduction in the size class of 100–799 beds. Cluster analysis identifies four clusters of provinces dominated by different hospital size classes after healthcare delivery reform in 2012. Cross-sectional time-series regression models find evolving impacts from market power, government intervention, and advancement of medical technology. We also offer a novel approach to understanding healthcare resource distribution by examining the structure rather than merely the number of healthcare resources.