Abstract. Most Earth System Models (ESMs) have incorporated, or are incorporating, coupled carbon and nutrient dynamics in their land modules. We show here that different numerical implementations of nutrient controls may imply different ecological mechanisms not recognized in the original model design and can have first order impacts on predicted terrestrial carbon cycling. Using ALM, the land module of the DOE ESM (ACME), we analysed land-atmosphere CO₂ exchange with coupled carbon and nitrogen dynamics through three commonly-applied numerical implementations of nitrogen limitation: (1) Mineral Nitrogen based Limitation (MNL), (2) Net nitrogen Uptake based Limitation (NUL), and (3) Proportional Nitrogen flux based Limitation (PNL). By the last decade of the contemporary period (1850–2000), the three schemes resulted in very similar global terrestrial carbon and nitrogen distributions. However, under an RCP4.5 CO₂ concentration forcing, the implementations resulted in wildly diverging 2001–2300 land-atmosphere CO₂ exchanges. Quantitatively, the divergence is as large as that of the CMIP5 models by 2100 and is about 1900 Pg C (~ 890 ppmv) by 2300. Our analysis suggests that these differences result from: (1) the typically high predicted terrestrial ecosystem carbon to nitrogen ratios (i.e., nutrient constrained conditions) and (2) the schemes predict different levels of nitrogen limitation to the carbon cycle, so that the PNL scheme leads to larger nitrogen loss through aerobic and anaerobic denitrification and surface and subsurface hydrological transport. We also found significant sensitivity of model predictions to initial conditions and numerical time step size but insignificant sensitivity to the sequence of numerical oxygen and nitrogen limitation or the ordering of reaction and chemical transport. We conclude that inconsistencies in imposing nutrient limitations will very likely lead to large uncertainties in predicted carbon stocks and long-term carbon-climate feedbacks. Finally, we recommend approaches to systematically alleviate these uncertainties.