Parametric heritage 3D reconstruction using iconographic evidence has the potential to extend modelling data sources and enhance the modelling efficiency. This study presents a novel and preliminary workflow to document, 3D reconstruct and interpret architectural massing typologies in the Song Dynasty of ancient China, integrating iconographic evidence. Multi-source historical images are collected and screened through chronological and appraisal assessments. The extracted massing principles are translated into parametric algorithms using Rhino-Grasshopper, then tested on 3D reconstruction of historical images by student participants. Furthermore, it is applied in an interactive design way through the Biomorpher plugin. The results show that the algorithms successfully enhanced the modelling efficiency, while limitations are also reflected. The interactive design application facilitates scenario comparisons in Song architecture 3D reconstruction and design. This study also sheds light on future applications in architectural conservation, interactive heritage education, and other research aspects in art and architecture.