The impact of climate change is felt globally and China is no exception. Evidence indicates that natural resource-based agriculture is especially sensitive and vulnerable to the adverse effects of climate change due to its strong dependence on weather conditions. Climate-related events such as droughts, cold waves, hailstorms, snowfall, erratic rainfall, and fluctuation in temperature contribute to farmers' vulnerability and affect the livelihood of farming households and agricultural production on which national food security depends. Under the background of global climate change, finding reasonable and effective adaptive measures to reduce agricultural losses and increase agricultural production levels have attracted increasingly more attention. Most studies pay attention to the relationship between farmers' perceptions and adaptive behaviors, while farmers' attributes are seldom considered. Examining the types of farmers' adaptive behaviors and the influencing mechanism among farmers' attributes, perception of climate change, and adaptive behaviors is of significance for making appropriate policies. Based on recent theoretical and empirical studies in the rain-fed agricultural areas, this study constructed a path analysis model with the perception of climate change as mediator to explore the path of impact among the variables of farmers' attributes, perception of climate change, and adaptive behaviors. The results show that: 1) The surveyed farmers showed a certain degree of recognition of local climate change and some actions adapting to climate change were taken. Among these, farmers prioritize technical measures, followed by economic measures, and few farmers adopted migration as an adaptive behavior. 2) The difference in the impact of farmers' attributes on adaptive behaviors is significant. 3) According to the mediating effects of the mediator, the model of multivariate path analysis was decomposed into three types: complete mediation model, partial mediation model, and no mediation model. Through the analysis of the above three models, it was found that farmers' attributes had an indirect impact on adaptive behaviors through mediating variables, as well as a direct impact through non-mediating variables. It indicates that the choice of adaptive behaviors depended not only on the farmers' perception of climate change, but also on their attributes. Finally, relevant policy recommendations were put forward with the aim to improve farmers' ability to cope with climate change.