Accurate quantification of methane (CH<sub>4</sub>) emissions from countries/regions is essential for effective emission reduction policy, but large uncertainties persist due to sparse observations and divergent bottom-up inventories. This study establishes a top-down Asian regional methane emission dataset based on the Methane Inversion inter-Comparison for Asia (MICA), using an ensemble of seven inversion systems (five global, two regional) assimilating surface in-situ measurements and GOSAT satellite retrievals for 2010–2021 under a common protocol. Validation against independent surface, aircraft, and balloon data demonstrates substantial improvements in model performance after inversion, with significant reductions in model-observation Root Mean Square Error (RMSE). Posterior estimates indicate an overall reduction of prior emissions by ~15% for the study area, with the largest decline in East Asia (~25.8%). Though a systematic downward adjustment by the inversion ensemble is observed for most top emitting-countries, the median differences between posterior and prior emissions vary across countries, from -0.9% (Mongolia) to -16.2% (China). Sectoral results suggest increasing emissions from waste management across developing Asian countries and a fourfold increase in Indonesian coal-mine emissions. Comparisons with National Greenhouse Gas Inventories (NGHGIs) reveal systematic underreporting by several nations — notably India, Pakistan, and Bangladesh. We find that inter-model spread is driven largely by horizontal resolution (Coefficient of Variation 10–20%), and that joint assimilation of satellite and surface observations is an effective option for constraining emissions in regions with sparse monitoring or complex land–sea geometries. This dataset provides a foundation for regional and national methane budget assessment in Asia and inverse model development, with robust consistency across multiple inversion systems and independent observations. The dataset generated in this study is openly available via Zenodo (<a href="https://doi.org/10.5281/zenodo.22699291">https://doi.org/10.5281/zenodo.22699291</a>, Wang et al., 2026).