The 58–51 Ma interval was characterized by a long-term increase of global temperatures (+4 to +6 • C) up to the Early Eocene Climate Optimum (EECO, 52.9– 50.7 Ma), the warmest interval of the Cenozoic. It was recently suggested that sustained high atmospheric pCO 2 , controlling warm early Cenozoic climate, may have been released during Neo-Tethys closure through the subduction of large amounts of pelagic carbonates and their recycling as CO 2 at arc volcanoes. To analyze the impact of Neo-Tethys closure on early Cenozoic warming, we have mod-eled the volume of subducted sediments and the amount of CO 2 emitted along the northern Tethys margin. The impact of calculated CO 2 fluxes on global temperature during the early Cenozoic have then been tested using a climate carbon cycle model (GEOCLIM). We show that CO 2 production may have reached up to 1.55 × 10 18 mol Ma −1 specifically during the EECO, ∼ 4 to 37 % higher that the modern global volcanic CO 2 output, owing to a dramatic India-Asia plate convergence increase. The subduction of thick Greater Indian continental margin carbonate sediments at ∼ 55–50 Ma may also have led to additional CO 2 production of 3.35 × 10 18 mol Ma −1 during the EECO, making a total of 85 % of the global volcanic CO 2 outgassed. However , climate modeling demonstrates that timing of maximum CO 2 release only partially fits with the EECO, and that corresponding maximum pCO 2 values (750 ppm) and surface warming (+2 • C) do not reach values inferred from geo-chemical proxies, a result consistent with conclusions arising from modeling based on other published CO 2 fluxes. These results demonstrate that CO 2 derived from decarbonation of Neo-Tethyan lithosphere may have possibly contributed to, but certainly cannot account alone for early Cenozoic warming. Other commonly cited sources of excess CO 2 such as enhanced igneous province volcanism also appear to be up to 1 order of magnitude below fluxes required by the model to fit with proxy data of pCO 2 and temperature at that time. An alternate explanation may be that CO 2 consumption, a key parameter of the long-term atmospheric pCO 2 balance, may have been lower than suggested by modeling. These results call for a better calibration of early Cenozoic weathering rates.