Heterogeneous chemistry and aerosol-radiation interaction (ARI) are considered two of the key mechanisms affecting haze formation and their reasonable parameterizations in atmospherically chemical models play an important role in accurate haze or air quality prediction. However, until the present, their relative contributions to severe haze have rarely been evaluated. Using the double-way atmospherically chemical model GRAPES_Meso5.1/CUACE CW, the relative contributions of heterogeneous chemistry and ARI to PM_2.5 are investigated under different pollution levels in Middle-Eastern China (MEC) during December 2016. Study results show that the combined effects of heterogeneous chemistry and ARI generate 29.7% of total PM_2.5 mass in MEC of December 2016, of which ARI contributes 19.7%, much higher than heterogeneous chemistry (6.1%). However, heterogeneous chemistry is important for reasonable reproduction of the proportion of secondary inorganic aerosols (SIA) of PM_2.5, especially during heavy haze. Both the contributions of heterogeneous chemistry and ARI to PM_2.5 increase with the haze pollution aggravation but to different degrees. The contribution of heterogeneous chemistry to modeled PM_2.5 concentration rises from 4.0% on clean days to 6.6% under moderate pollution and 7.8% under heavy haze. And that of ARI increases rapidly from 9.1% on clean days to 21.0% under moderate pollution and 29.1% during heavy pollution. It is worth noticing that there is an additional contribution to PM_2.5 caused by the combination of heterogeneous chemistry and ARI except for their respective contributions, which reaches up to 7.3% of PM_2.5 mass and 18.5% of SIA generation under heavily polluted conditions. This is because ARI further promotes heterogeneous oxidation of SO₂ and NO_x by increasing the aerosol surface area concentration (SAC), causing additional SIA production via heterogeneous pathways. This work indicates the contribution of ARI to severe haze formation not only by altering meteorological conditions but also by reforming the chemical processes in the model.