Freezing can notably alter the cutting behaviour of wood, yet quantifying its influence is challenging due to variability in wood properties and the interplay of factors such as moisture content (MC), temperature and density. This study uses a tightly controlled design to isolate the effect of freezing on cutting forces of Scots pine (Pinus sylvestris L.) in green state. Heartwood and sapwood were analysed separately due to their distinct MC, at four temperature levels: -20 ^∘ C, -10 ^∘ C, 0 ^∘ C, and 22 ^∘ C. A data-driven machine-learning approach, combined with SHAP analysis and partial dependence plots, was used to identify and interpret the most influential variables on main cutting force. Results show that heartwood at 30 0 and -10 ^∘ C as free water freezes. This phase change alters the cutting mechanism, reducing friction but increasing resistance when cutting through ice and the ice–wood composite, with the net effect depending on uncut chip thickness. These findings provide mechanistic insight into temperature-dependent cutting behaviour and inform optimisation of sawing strategies for frozen logs in industrial practice.