Abstract Climate warming will affect terrestrial ecosystems in many ways, and warming‐induced changes in terrestrial carbon (C) cycling could accelerate or slow future warming. So far, warming experiments have shown a wide range of C flux responses, across and within biome types. However, past meta‐analyses of C flux responses have lacked sufficient sample size to discern relative responses for a given biome type. For instance grasslands contribute greatly to global terrestrial C fluxes, and to date grassland warming experiments provide the opportunity to evaluate concurrent responses of both plant and soil C fluxes. Here, we compiled data from 70 sites (in total 622 observations) to evaluate the response of C fluxes to experimental warming across three grassland types (cold, temperate, and semi‐arid), warming methods, and short (≤3 years) and longer‐term (>3 years) experiment lengths. Overall, our meta‐analysis revealed that experimental warming stimulated C fluxes in grassland ecosystems with regard to both plant production (e.g., net primary productivity (NPP) 15.4%; aboveground NPP (ANPP) by 7.6%, belowground NPP (BNPP) by 11.6%) and soil respiration (Rs) (9.5%). However, the magnitude of C flux stimulation varied significantly across cold, temperate and semi‐arid grasslands, in that responses for most C fluxes were larger in cold than temperate or semi‐arid ecosystems. In semi‐arid and temperate grasslands, ecosystem respiration (Reco) was more sensitive to warming than gross primary productivity (GPP), while the opposite was observed for cold grasslands, where warming produced a net increase in whole‐ecosystem C storage. However, the stimulatory effect of warming on ANPP and Rs observed in short‐term studies (≤3 years) in both cold and temperate grasslands disappeared in longer‐term experiments (>3 years). These results highlight the importance of conducting long‐term warming experiments, and in examining responses across a wide range of climate.