Abstract Lakes (including reservoirs) are an important component of the global carbon (C) cycle, as acknowledged by the fifth assessment report of the IPCC . In the context of lakes, the boreal region is disproportionately important contributing to 27% of the worldwide lake area, despite representing just 14% of global land surface area. In this study, we used a statistical approach to derive a prediction equation for the partial pressure of CO 2 ( p CO 2 ) in lakes as a function of lake area, terrestrial net primary productivity ( NPP ), and precipitation ( r 2 = .56), and to create the first high‐resolution, circumboreal map (0.5°) of lake p CO 2 . The map of p CO 2 was combined with lake area from the recently published GLOWABO database and three different estimates of the gas transfer velocity k to produce a resulting map of CO 2 evasion ( F CO 2 ). For the boreal region, we estimate an average, lake area weighted, p CO 2 of 966 (678–1,325) μatm and a total F CO 2 of 189 (74–347) Tg C year −1 , and evaluate the corresponding uncertainties based on Monte Carlo simulation. Our estimate of F CO 2 is approximately twofold greater than previous estimates, as a result of methodological and data source differences. We use our results along with published estimates of the other C fluxes through inland waters to derive a C budget for the boreal region, and find that F CO 2 from lakes is the most significant flux of the land‐ocean aquatic continuum, and of a similar magnitude as emissions from forest fires. Using the model and applying it to spatially resolved projections of terrestrial NPP and precipitation while keeping everything else constant, we predict a 107% increase in boreal lake F CO 2 under emission scenario RCP 8.5 by 2100. Our projections are largely driven by increases in terrestrial NPP over the same period, showing the very close connection between the terrestrial and aquatic C cycle.