The primary study objective is to verify the environmental Kuznets curve hypothesis for greenhouse gas emissions in Poland at the disaggregated Nomenclature of Territorial Units for Statistics (NUTS)-3 level, accounting for spatial dependencies. The analysis employs a balanced panel of 73 subregions (2005 to 2022) based on the Emissions Database for Global Atmospheric Research and Statistics Poland data. Spatial dependencies were determined using Moran’s and Lagrange multiplier tests, justifying the application of a spatial autoregressive panel model with individual fixed effects. The results offer robust evidence of an inverted U-shaped relationship between economic growth and emissions. The estimated turning point is about 73,035 PLN (16,592 USD) at the NUTS-3 level in 2022 prices. Urbanisation exerts a positive, statistically significant influence on per capita emissions, whereas household energy consumption indicates a positive, weakly significant influence (p < 0.10). Robustness checks performed at NUTS-2 confirmed the findings, yielding a turning point of 59,704 PLN (13,564 USD), although energy consumption proved statistically insignificant at this aggregated scale. These findings highlight the importance of incorporating spatial spillovers, as neighbouring dynamics shape regional emissions. The results underscore the need for differentiated regional climate policies reflecting spatial and structural disparities across Polish regions.