Purpose: The major limitation of tumor microwave ablation (MWA) operation is the lack of predictability of the ablation zone before surgery. Operators rely on their individual experience to select a treatment plan, which is prone to either inadequate or excessive ablation. This paper aims to establish an ablation prediction model that guides MWA tumor surgical planning. Methods: An MWA process was first simulated by incorporating electromagnetic radiation equations, thermal equations, and optimized biological tissue parameters (dynamic dielectric and thermophysical parameters). The temperature distributions (the short/long diameters, and the total volume of the ablation zone) were then generated and verified by 60 cases ex vivo porcine liver experiments. Subsequently, a series of data were obtained from the simulated temperature distributions and to further fit the novel ablation coagulated area prediction model (ACAPM), thus rendering the ablation-dose table for the guiding surgical plan. The MWA clinical patient data and clinical devices suggested data were used to validate the accuracy and practicability of the established predicted model. Results: The 60 cases ex vivo porcine liver experiments demonstrated the accuracy of the simulated temperature distributions. Compared to traditional simulation methods, our approach reduces the long-diameter error of the ablation zone from 1.1$$cm to 0.29$$cm, achieving a 74% reduction in error. Further, the clinical data including the patients’ operation results and devices provided values were consistent well with our predicated data, indicating the great potential of ACAPM to assist preoperative planning.