The incomplete combustion of vegetation and dead organic matter by landscape fires creates recalcitrant pyrogenic carbon (PyC), which could be consequential for the global carbon budget if changes in fire regime, climate, and atmospheric CO2 were to substantially affect gains and losses of PyC on land and in oceans. Here, we included global PyC cycling in a coupled climate–carbon model to assess the role of PyC in historical and future simulations, accounting for uncertainties through five sets of parameter estimates. We obtained year-2000 global stocks of (Central estimate, likely uncertainty range in parentheses) 86 (11-154), 47 (2-64), and 1129 (90-5892) Pg C for terrestrial residual PyC (RPyC), marine dissolved PyC, and marine particulate PyC, respectively. PyC cycling decreased atmospheric CO2 only slightly between 1751 and 2000 (by 0.8 Pg C for the Central estimate) as PyC-related fluxes changed little over the period. For 2000 to 2300, we combined Representative Concentration Pathways (RCPs) 4.5 and 8.5 with stable or continuously increasing future fire frequencies. For the increasing future fire regime, the production of new RPyC generally outpaced the warming-induced accelerated loss of existing RPyC, so that PyC cycling decreased atmospheric CO2 between 2000 and 2300 for most estimates (by 4-8 Pg C for Central). For the stable fire regime, however, PyC cycling usually increased atmospheric CO2 (by 1–9 Pg C for Central), and only the most extreme choice of parameters maximizing PyC production and minimizing PyC decomposition led to atmospheric CO2 decreases under RCPs 4.5 and 8.5 (by 5-8 Pg C). Our results suggest that PyC cycling will likely reduce the future increase in atmospheric CO2 if landscape fires become much more frequent; however, in the absence of a substantial increase in fire frequency, PyC cycling might contribute to, rather than mitigate, the future increase in atmospheric CO2.