A major challenge in biogeochemistry is to reduce the uncertainty of projections made by soil carbon models. In the last two decades, carbon-use efficiency, the proportion of consumed carbon incorporated into microbial biomass, has become a central parameter to represent microbial control on carbon fluxes. For models that integrate other elements, like nitrogen, the adjustment of carbon-use efficiency in response to substrate stoichiometry has gained popularity as a mechanism to balance these fluxes, mostly due to its mathematical convenience. The reasoning behind this, is that microbes release the excess carbon as CO<sub>2</sub>, a mechanism known as overflow respiration. This mechanism, however, causes a characteristic decrease of carbon-use efficiency when reaching nitrogen limitation. In this study we propose that the implementation of overflow respiration forces an unrealistic decrease of carbon-use efficiency for three reasons: 1) physiological mechanisms can minimize carbon excess, avoiding overflow; 2) carbon overflow has been reported in laboratory experiments mainly as dissolved organic carbon (organic acids), and not as CO<sub>2</sub>; 3) functionally diverse microbial communities can exhibit higher-level dynamics, improving the recycling of nutrients and avoiding overflow. We use an individual-based microbial litter decomposition model to test the impact of these mechanisms on carbon-use efficiency. We found that physiological mechanisms such as flexible biomass stoichiometry can eliminate overflow, but nutrient allocation does not. When carbon overflow occurs as dissolved organic carbon, carbon-use efficiency increases under nitrogen limitation. Finally, a functionally diverse community can avoid carbon overflow, although carbon-use efficiency declines due to higher maintenance respiration. We demonstrate that the representation of overflow respiration in soil carbon models is more relevant than currently acknowledged. Redirecting carbon overflow to a dissolved organic carbon pool can lead to opposite trends in carbon losses. The soil carbon modeling community should thoroughly assess current implementations and explore more mechanistically grounded alternatives. This includes dissolved organic carbon pathways, more realistic microbial community representations, or other processes that better capture the complexity of microbial carbon use.