The Yoshino cherry tree (Cerasus × yedoensis ‘Somei-yoshino’) is a spring symbol in Japan. For the cherry tree to bloom, dormant buds must experience a cold exposure period to break their endo-dormancy. Unlike flowering, bud endo-dormancy release is not visually detectable. Although DORMANCY-ASSOCIATED MADS-box (DAM) genes are known to play an important role in this process, their specific functions in natural environments remain unclear. To develop a genetically informed mathematical model for predicting endo-dormancy release, we examined seasonal changes in the genome-wide gene expression profiles of Yoshino cherry trees at three sites in Japan along a latitudinal gradient: Sapporo, Tsukuba, and Fukuoka. We identified five transcriptional modes using gene expression analysis and classified them into seasonal categories: early summer, summer, autumn, winter, and spring. By comparing these modes with daily mean temperatures, we found that the winter and spring profiles emerged when the daily mean temperature dropped to approximately 10°C. Among the six DAM genes, DAM4 expression most closely reflected the rate of endo-dormancy release. DAM4 expression increases in early winter but is subsequently decreased during the colder winter months. Our results suggest that Yoshino cherry tree requires exposure to temperatures below approximately 10.3°C in Sapporo, 8.7°C in Tsukuba, and 11.3°C in Fukuoka for 61.1 days to sufficiently suppress DAM4 gene expression, reaching the critical threshold for endo-dormancy release. According to this prediction, endo-dormancy release was estimated to be delayed by 7.3 days in Sapporo, 6.6 days in Tsukuba, and 6.4 days in Fukuoka from 1990 (the mean date from 1961 to 1990) to 2020 (the mean date from 1991 to 2020). This analysis indicates that endo-dormancy release was delayed by approximately 2.3 days per decade from 1990 to 2020. These findings demonstrate that gene expression, particularly that of DAM4, serves as a reliable indicator for predicting the timing of endo-dormancy release. Also, the predictive model for endo-dormancy break developed in this study is expected to be applicable to other Rosaceae species possessing DAM genes and will provide valuable insights for addressing the effects of climate change. To improve the model’s accuracy, future studies should conduct sampling 2–4 times per month to refine dormancy release predictions based on gene expression. This study, conducted by Atsuko Miyawaki-Kuwakado et al., was published in Plants, People, Planet (2024).