Methodologies for the construction of nominal and real gross domestic product (GDP) time series often differ over time and between countries. This paper discusses the main issues raised by this methodological heterogeneity for long-run measures of economic growth and, informed by these issues, provides a set of internationally comparable GDP estimates from 1820 to 2020. The estimates are based on real product benchmarks relative to the United Kingdom as the reference economy. The GDP time series of the reference economy is a normalized composite of several indices. These estimates suggest that the Maddison Project data sets overestimate economic growth after 1950 relative to the period from 1820 to 1950.