<p>The urban heat island effect poses a serious threat to human health and urban sustainability, with extreme heat (EH) events exacerbating its severity. However, there is a lack of research on the response of regional heat islands (RHI) under EH scenarios compared to normal weather (NW). This study utilized daily surface temperature datasets to analyze the spatial distribution characteristics of RHI in the Zhengzhou metropolitan agglomeration under EH and NW. Additionally, interpretable machine learning models were employed to explore the relative importance and interactions of natural and social factors influencing RHI in these scenarios. The results indicate that (1) compared to NW, the area of RHI under EH expanded by 1.1 to 1.3 times, forming highly intensive, cross-city aggregated RHIs. (2) During the day, RHI was mainly influenced by urban-rural evaporation, with ecological pattern factors (PEL, AI) being the key drivers, exhibiting stronger effects under EH. At night, anthropogenic heat emissions and surface heat storage were the main sources of RHI, with human activity factors (NTL) and urban construction factors (AH, FAR) being significant indicators, having a higher proportion under EH. (3) Their interaction indicated that human activity and urban construction elements weaken the cooling effect of ecological pattern, while the synergistic effect of ecological pattern factors helped to form high-quality regional cooling sources. (4) Based on these findings, a heat risk zoning control strategy was proposed. This study supports the mitigation of potential synergistic effects between RHI and EH and the sustainable development of regional thermal environments.</p>