The increasing frequency and severity of extreme air temperature events (frost and heat) due to climate change significantly reduce crop yield worldwide. Crop modelling has been widely used to quantify the impacts of climate change on crop productivity. However, most crop models, including APSIM-Canola and APSIM-Wheat, capture continuous temperature responses by adjusting phenology, photosynthesis, and leaf growth and senescence, but lack explicit mechanisms to quantify short and event-based frost and heat damage to yield. This study developed and integrated frost and heat damage functions into APSIM to estimate yield reductions based on daily air temperature and the sensitivity of the crop growth stage. We applied a combined process-based modelling and data-driven approach to parameterise the damage functions using extensive historical field data from the Australian grain belt. Crop-specific parameters (Stevenson‑screen air temperature thresholds, sensitive periods, and maximum daily yield losses) were identified for frost and heat stress via differential evolution within literature‑informed bounds and subsequently validated against a season‑level hold-out testing set. The inclusion of the parameterised damage functions improved the overall model performance of yield predictions compared with the original APSIM for the examined Australian datasets. The overall RRMSE decreased by 13.7% for canola and 80.5% for wheat in the testing sets. These improvements were consistently observed across the examined datasets for both crops under the season‑level hold‑out evaluation. Integrating these functions with specific APSIM crop models is expected to improve the ability of APSIM to simulate yield under comparable conditions, improve identification of optimal flowering periods, and facilitate modelling analyses of genotype, environment, and management interactions. Access to more comprehensive and diverse field datasets will enable further refinements of these functions in the future.