Predicting tourism flow networks offers important insights for destination development but remains a methodological challenge. This study develops a novel steady-state Markov chain method, leveraging trajectory big data to facilitate short- and long-term predictions of interactions and distributions between nodes in tourism flow networks, using Tibet as a case study. The results demonstrate that the tourism flow network effectively elucidates intricate interrelationships. In the short term, the one-step transition probability matrix identifies tourists’ potential next destinations, reflecting dynamic network changes. Over the long term, the Markov chain steadystate vector uncovers the stable distribution of tourists, emphasizing shifts in node significance. Additionally, the determinants influencing tourism destinations and different nodes have continuously evolved, whether assessed from a global influence or spatial heterogeneity perspective. Beyond its theoretical contributions, this paper offers practical implications for destination planning and intelligent decision-making management information systems.