Radiocarbon dates have become a cornerstone in archaeological reconstructions of past population dynamics. The increasing reliance on large-scale radiocarbon databases, usually aggregated from diverse sources, however, raises persistent concerns about sampling bias, especially heterogeneous sampling intensity across sites. In this paper, we introduce a rescaling method that adjusts the frequency of dates in radiocarbon datasets in proportion to dwelling counts at the settlement level, using weighting and bootstrap resampling. Through a series of simulations, we show that this approach consistently yields probability distributions that more closely reflect hypothetical population trends, particularly in contexts with high inter-settlement variability in sampling intensity. We apply our method to archaeological data from two areas in Korea, the Yeongsan and Geum River Basins, during the Proto–Three Kingdoms (1C BC–AD 3C) and Three Kingdoms Periods (AD 4–7C). Results demonstrate that rescaled datasets offer significantly different interpretations of population organization and reconfiguration than those derived from original data. This study highlights the importance of addressing sampling heterogeneity in local-scale demographic research and suggests that rescaling is a valuable complement to existing bias-correction strategies in archaeological studies of demography.