Caffeine, a typical representative of methylxanthine, is world-widely used to manage apnea of prematurity (AOP) in neonatology. However, an appropriate medication regimen of caffeine has not been well defined until now. For example, in terms of the duration of caffeine, AAP guideline for AOP (2016) and British NICE guideline for neonatal respiratory care (2019) all recommended discontinuing caffeine when the infants reached a postmenstrual age (PMA) ≥33weeks and had a stable respiratory status, commonly manifested by weaning from non-invasive ventilation and free of apneic episodes for at least five consecutive days. Interestingly, the actual clinical settings seem to be not strictly following this recommendation. A survey of the neonatologist in North America revealed that a substantial variability existed among sites in the timing of caffeine discontinuation before discharge and the respiratory support at the time of caffeine discontinuation \[1\]. Another survey in Saudi Arabia also had a similar finding \[2\]. The optimal timing of discontinuing caffeine is still a conundrum in the field of neonatology. Ideally, the optimal timing of discontinuing caffeine should be individual-specific. Published work has indicated that AOP and intermittent hypoxemia (IH) were frequently observed beyond 36 weeks' PMA in all gestational age groups, particularly in the 24- to 27-week infants \[3, 4\]. In the clinical settings, intermittent hypoxic and AOP episodes is a predominant cause of oxygen supplement in premature infants and commonly prolong the hospital stay. Optimizing arterial saturation by oxygen supplement is essential to achieve a stable cardiorespiratory status because hypoxemia could induce hypoxic sensitivity of the carotid bodies in neonates, resulting in more pronounced ventilatory depression and more frequent apneic episodes. Some RCTs have shown that continuing caffeine administration beyond PMA 34 weeks could reduce the frequency of IH episodes in premature infants \[4, 5\]. Therefore, theoretically, a prolonged caffeine administration over the usual duration could shorten the duration of oxygen supplements in those infants at high risk of frequent late AOP or IH. Target weaning oxygen could be an opportunistic indicator of discontinuing caffeine. In light of the above considerations, a multicenter, retrospective, partially blinded, controlled trials will be conducted to verify the hypothesis that a novel caffeine regimen that weaning oxygen as the indicator of discontinuing caffeine could improve respiratory outcomes of very premature infants.