[Objective] As agricultural modernization accelerates, intelligent agricultural machinery is playing an increasingly important role in enhancing productivity and ensuring food security. Among these technologies, BeiDou navigation tractors have shown considerable potential in improving operational precision and efficiency. However, their adoption among farmers remains limited, and large-scale diffusion has yet to be achieved. Understanding what drives or constrains farmers' adoption decisions is therefore critical for promoting the effective use of intelligent agricultural machinery. The aim of this research is to identify the key determinants of farmers' adoption behavior and to uncover the underlying mechanisms shaping these decisions. [Methods] Micro-survey data collected from 1 242 maize farmers across six provinces in China were used as the basis. An analytical framework was constructed from five dimensions, including household head characteristics, family endowments, operational conditions, regional development, and the external environment. Explainable machine learning methods were employed to identify the key driving factors influencing farmers' adoption of BeiDou navigation tractors. [Results and Discussions] The results showed that farmers' adoption decisions regarding BeiDou navigation tractors were characterized by the dominance of a small number of core factors. Among these, policy support for intelligent agricultural machinery, total household income, farm size, age, and participation in agricultural training ranked as the top five determinants. This indicated that policy incentives, economic capacity, land endowments, and human capital were the primary factors influencing farmers' adoption decisions. In terms of the direction of influence, stronger human, physical, and social capital effectively reduced the barriers to technology adoption. The consolidation and contiguity of farmland, together with the application of complementary technologies, significantly improved the suitability of BeiDou navigation tractors. In addition, a well-developed policy support system and agricultural socialized service network created favourable conditions for adoption. The heterogeneity analysis further revealed that smallholder farmers relied more on demonstration effects and external service support, whereas large-scale farmers' adoption decisions were more strongly driven by policy incentives and economic returns. Younger farmers exhibited adoption behaviour that was more evidently driven by economic and policy factors, while older farmers showed greater dependence on services and a stronger tendency towards risk aversion. Farmers operating under less favourable terrain conditions were more influenced by resource endowments and policy support, whereas those in more favourable conditions were more driven by access to services and human capital. [Conclusions] Based on these findings, it is suggested that promoting the large-scale adoption of intelligent agricultural machinery requires improvements in policy support. First, the focus should shift from hardware subsidies to performance enhancement. On the one hand, differentiated subsidy policies should be implemented. On the other hand, subsidy mechanisms should be extended from equipment purchase to operational use. Second, efforts should be made to promote moderate-scale farmland management in order to improve the conditions for applying intelligent agricultural machinery. This includes facilitating land transfer and service arrangements such as land trusteeship to encourage land consolidation and contiguity, thereby enabling the precision operation advantages of BeiDou navigation tractors to be fully realised. At the same time, the coordinated application of agricultural machinery and agronomic practices should be promoted. Under suitable conditions, integrating BeiDou navigation tractors with high-density planting techniques can further enhance their cost-saving and efficiency-improving potential. Finally, it is necessary to improve the agricultural socialized service system and training framework, and to implement differentiated promotion strategies for different groups. On the one hand, greater support should be given to agricultural machinery cooperatives and agricultural service organisations to provide outsourced services such as ploughing, planting, and land management for smallholders. On the other hand, more digital skills and practical training resources should be directed towards large-scale farmers, alongside technical guidance on equipment maintenance and data management, so as to better utilise their leading role in the adoption and diffusion of intelligent agricultural machinery.