Objective. The objective of this study was to determine personalized optimal timing for re-planning in adaptive organ-at-risk-sparing radiation therapy under limited re-planning resources in patients with head and neck cancer (HNC). Approach. A novel Markov decision process (MDP) model was developed to determine optimal timing of re-plannings based on the patients expected toxicity, characterized by normal tissue complication probability (NTCP), for four toxicities: xerostomia, dysphagia, parotid gland dysfunction, and feeding tube dependency at 6 months post-treatment. The MDP parameters were derived from a dataset comprising 52 HNC patients treated at the University of Texas MD Anderson Cancer Center between 2007 and 2013. Optimal re-planning strategies were obtained when the permissible number of re-plannings throughout the treatment was limited to 1, 2, and 3. Main results. Based on the MDP solution, it is optimal to re-plan when the difference between planned and actual NTCPs (ΔNTCP) was greater than or equal to 1%, 2%, 2%, and 4% at treatment fractions 10, 15, 20, and 25, respectively, exhibiting a temporally increasing pattern. The ΔNTCP thresholds remained constant across the number of re-planning allowances (1, 2, and 3). This result underscores the importance of re-planning for patients experiencing the slightest change in ΔNTCP at fraction 10. Significance. In contrast to prior work that relies on a single re-planning allowance or predetermined time intervals using a one-size-fits-all approach, the MDP model proposed in this paper offers a personalized, resource-aware, and scalable decision-making tool; it identifies optimal dynamic re-planning schedules tailored to individual needs, guided by changes in NTCP.