In this chapter I discuss historical estimates of GDP at both the national and the regional level, and their application for assessing economic performance in modern times. Having been invented in (and conceived for) industrial capitalist societies, GDP has stronger informative power in those contexts where industry and services, and market exchange, retain the lion's share of production. In modern times, when comparing the series available for different countries, there are three major methodological problems to be acknowledged and possibly addressed: the dissimilarity of the quantity series and related proxies; deflation through purchasing power parities distant in time; and the differences in the base year used to construct GDP constant price (Laspeyres) indices (the latter issue may be less widely recognized, but it may have a remarkable impact). The way the estimates are constructed also has a bearing upon the statistical tools and models we should use to interpret them; owing to the lack of reliable long-run series, cross-sectional techniques are often preferable to time series analysis; provided we have reliable estimates, growth accounting − decomposing GDP growth into productivity and industry mix effects − may provide important clues about the choice between theoretical approaches; not least for the quality of our data, crosscountry convergence models based on conditioning variables should always be supplemented by historical information from qualitative sources and case studies. More generally, cliometricians should prove themselves capable of adapting their models to different historical contexts and relativizing findings to the limits of their estimates.