<p>Urban clusters have become the fundamental spatial units for achieving carbon reduction, yet the pivotal interactive role of urban-scale and urban cluster-scale population agglomeration in shaping carbon emission intensity (CEI) has been largely overlooked. To address this gap, we develop a dual-scale analytical framework to examine how urban-scale population agglomeration affects CEI and how this relationship is moderated by urban cluster-scale population agglomeration. Based on panel data for 221 cities in 19 urban clusters in China from 2011 to 2022, we employ a two-way fixed effects model for the empirical analysis. The results reveal a robust U-shaped relationship between urban-scale population agglomeration and CEI: urban-scale population agglomeration initially reduces CEI but increases it beyond a turning point. More importantly, urban cluster-scale population agglomeration, measured using an inverse-distance-weighted agglomeration index, significantly flattens this U-shaped relationship. Green technological innovation and the co-agglomeration of diverse functional segments are identified as the two main channels through which this moderating effect operates. Heterogeneity analyses show that the moderating effect is more pronounced in eastern coastal and polycentric urban clusters. Moreover, this effect is stronger within a limited geographic range. These findings provide actionable insights for designing coordinated population agglomeration strategies at both the urban and urban cluster scales and for promoting carbon reduction across multiple scales within urban clusters.</p>