The multi-risk assessment of agrometeorological disasters is of great significance for disaster prevention and mitigation and the development of risk countermeasures. According to the annual statistics of agrometeorological disasters in Shaanxi province from 1980 to 2015 and the survey data of farmers in 2015-2018, we use multi-risk assessment methods such as Bayesian network and Bayesian hierarchical model (BHM), analyze the change trend of agrometeorological disasters in the province, as well as the interactions between disasters and the impact of livelihood capital on the vulnerability of different crops. The results show that: (1) The rate and intensity of flood disasters in Shaanxi Province from 1980 to 2015 dropped significantly, with slopes of 0.17% and 0.7%, respectively, and the disaster rate of frosts increased significantly, with a slope of 0.25%, while there was no significant change in drought and hail disasters. (2) Droughts and floods significantly aggravated other disasters, and there were complex interactions between various disasters. (3) When we reduced vulnerability by increasing livelihood capital, these livelihood capital indicators, such as the ratio of effective irrigated area, crop planting area, rural electricity consumption and agricultural phosphate fertilizer application, should be selected. The study will improve the adaptability of farmers to agrometeorological disasters and provide guidance for the prevention and mitigation of agrometeorological disasters.