This paper assesses the impact of air–sea coupling on Indian Ocean tropical cyclones (TCs) by comparing a 20-year long simulation of a ¼° regional coupled ocean–atmosphere model with a twin experiment, where the atmospheric component is forced by sea surface temperature from the coupled simulation. The coupled simulation reproduces the observed spatio-temporal TCs distribution and TC-induced surface cooling reasonably well, but overestimates the number of TCs. Air–sea coupling does not affect the cyclogenesis spatial distribution but reduces the number of TCs by ~ 20% and yields a better-resolved bimodal seasonal distribution in the northern hemisphere. Coupling also affects intensity distribution, inducing a four-fold decrease in the proportion of intense TCs (Cat-2 and stronger). Air–sea coupling damps TCs growth through a reduction of inner-core upward enthalpy fluxes due to the TC-induced cooling. This reduction is particularly large for the most intense TCs of the northern Indian Ocean (up to 250 W m−2), due to higher ambient surface temperatures and larger TC-induced cooling there. The negative feedback of air–sea coupling on strongest TCs is mainly associated with slow-moving storms, which spend more time over the cold wake they induce. Sensitivity experiments using a different convective parameterization yield qualitatively similar results, with a larger (~ 65%) reduction in the number of TCs. Because of their relatively coarse resolution (¼°), both set of experiments however fail to reproduce the most intense observed TCs. Further studies with finer resolution models in the Bay of Bengal will be needed to assess the expectedly large impact of air–sea coupling on those intense and deadly TCs.