Objective To construct and validate a predictive model for the risk of bone metastasis in patients with primary liver cancer. Methods The research was generally divided into two parts: model establishment and model validation. A total of 197 patients with primary liver cancer from January 2018 to June 2018 were selected to be included in the study when building the model, and the nomogram prediction model based on Cox regression was used in the case-control study method. The validation process continued to select 238 patients with primary liver cancer (no bone metastasis) in our hospital (from July 2018 to December 2018) and followed up for 3 years. The information of the prognosis of bone metastasis during the follow-up period was observed and collected to complete the validation. SPSS statistical software and R software were used to complete the data analysis. Results The results of regression analysis at the stage of building the model showed that age, family history of malignant tumor, previous history of hepatitis B, tumor stage, primary focus surgery and tumor differentiation were independent factors affecting the prognosis of patients with bone metastasis (P <0.05). The nomogram clinical prediction model was established by using R software. The prediction model finally included four factors: age, previous history of hepatitis B, primary surgery, and degree of differentiation. The AUC of ROC curve for predicting the risk of bone metastasis was 0.758. Conclusion The nomogram model constructed in this study has a medium to high degree of predictive calibration for predicting the risk of bone metastasis in patients with primary liver cancer within 3 years and is worthy of clinical auxiliary use.