Oil recovery, power generation, water desalination, gas flaring, and traffic are the main contributors to SO2 emissions in the Middle East (ME). Satellite observations suggest that the traditional emission inventories do not account for multiple SO2 emission sources in the ME. This study aims to evaluate the most frequently used SO2 emissions datasets over the ME by comparing high-resolution regional model simulations and meteorology/chemistry assimilation products, MERRA-2 and CAMS, with satellite and available ground-based air-quality observations. Here, we employ the WRF-Chem-3.7.1 regional meteorology-chemistry model and conduct simulations for the period 2015-2016 with 10 km grid spacing using HTAP-2.2 emission datasets and the new OMI-HTAP data, which is based on the combination of the near-surface SO2 emissions taken from the HTAP-2.2 inventory with strong (>30 kt/year) SO2 point sources obtained from the satellite Ozone Monitoring Instrument (OMI) observations. We find that conventional emission inventories (EDGAR-4.2, MACCity, and HTAP-2.2) have uncertainties in the location and magnitude of SO2 sources in the ME and significantly underestimate SO2 emissions in the Arabian Gulf. The WRF-Chem, run in conjunction with the new OMI-HTAP emissions, improves comparisons between the satellite and ground-based SO2 observations. Our simulations show that SO2 surface concentrations in Jeddah and Riyadh frequently exceed European air-quality limits. The ME generates about 10% of global anthropogenic SO2 emissions, on par with India. Therefore, the development of effective emission controls and improvement of air-quality monitoring in the ME are urgently needed.