Key message High starch concentrations in woody tissue of sweet cherry trees were associated with deep stages of dormancy, while increasing concentrations of hexoses were related to the dormancy release process. Abstract Dormancy is an intrinsic characteristic of deciduous forest and fruit trees. Release from dormancy and subsequent bud burst occur after chill requirements (CR) have been met. Currently, chilling metrics are poorly interchangeable among climates, probably due to a lack of physiological parameters involved in earlier model development, and, therefore, the reported CR values differ across locations. Here, we propose a way to improve our understanding on physiological markers involved in dormancy of deciduous fruit trees studying the dynamics of starch and hexoses (glucose + fructose) in sweet cherry twigs. Probabilities of bud break were estimated through logistic regression analysis using as independent variable the concentration of carbohydrates. In this experiment, we used 960 twigs and 160 sub-samples of twig portions ( extasciitilde 2 cm of stem) from 8 sweet cherry cultivars exposed to different amounts of chill in the field. The timing of bud burst in forcing conditions and the concentration of starch and hexoses in the sub-samples were recorded. We found that concentrations of starch and hexoses are closely related to dormancy progression. Specifically, starch concentrations decreased by 43%, while concentrations of hexoses increased by 58% during the transition from 16 to 70 Chill Portions. Responses differed among varieties at the same chill received in the orchard. The logistic regression analysis revealed that the predicted probability of bud break at the moment of recorded bud burst was highest (i.e., up to 0.5) in the varieties Santina, Skeena, Rainier, Lapins, Regina, Kordia and Bing. In contrast, the variety Sweetheart showed the lowest probability (i.e., 0.49) of having met its CR when bud burst was recorded. This may suggest that in this variety, carbohydrate metabolism is not the most important predictor involved in dormancy release. The method presented here improves our understanding on dormancy and might serve to focus further research on dormancy modeling. This is especially important in warm winter regions where the classical way of determining CRs and quantifying chill accumulation is consistently failing.