Global terrestrial ecosystems absorb about a third of anthropogenic emissions each year, due to the difference between two key processes: photosynthesis and respiration. Despite the importance of these two processes at the global scale, no direct measurement exists of either. Eddy-covariance (EC) measurements have been widely used as the closest ‘quasi-direct’ observation, and the resulting estimates have been used to produce global budgets of photosynthesis and respiration. Recent research, however, suggests that current estimates may be biased by up to 25%, as the methods used to partition observed net carbon fluxes to photosynthesis and respiration do not take into account any inhibition of leaf respiration in light. Yet the prevalence of light-inhibition of leaf respiration remains debated, and impacts on global estimates of photosynthesis and respiration unquantified. Here, we use novel approaches to estimate the extent of light-inhibition across the global FLUXNET EC network, and find strong evidence for an inhibition effect on ecosystem respiration, which varies by season and plant functional type. We develop partitioning methods that allow for inhibition, and find that that diurnal patterns of ecosystem respiration might be markedly different than previously thought. The results call for the reevaluation of global terrestrial carbon cycle models, and also suggest that current global budgets of photosynthesis and respiration may be biased on the order of magnitude of anthropogenic fossil fuel emissions.