Inclusion in civil registries is key for establishing an individual’s legal identity, and for accessing basic rights and services. From the state’s perspective, complete civil registries are crucial for rendering societies legible, and thus for public goods provision. Yet, according to the World Bank, an estimated 850 million people worldwide lack official identity documentation. Existing literature tends to attribute gaps in registration coverage to insufficient capacity, with low-income countries falling behind. Yet, there is considerable variation in the extent to which different types of regimes invest in rendering their populations legible and in building an infrastructure for registration. There is also unexplained variation in the extent to which citizens are willing to comply with the state’s demand for information. Several socialist or post-communist countries with otherwise low levels of development have achieved near-universal coverage, whereas democratic countries like India have struggled with registration. How do regime characteristics shape civil registration coverage? Which types of regimes are successful in rendering their populations legible? Drawing on a range of cross-national data sources such as rates of birth registration, national ID coverage, census-taking frequency, and age-heaping measures, this paper employs structural equation modelling (SEM) to explore the political determinants of civil registration coverage – conceptualised as an unobserved latent variable. SEM is used to examine a theoretical model of how political factors influence registration, which in turn manifests as variation in observable measures of identity documentation. This study contributes to the emerging cross-national literature on legibility and an important challenge in human development.