In the modern product development process, newly developed products have to be tested in terms of their analytical and sensorial stability throughout the whole shelf-life. A real-time storage at ambient conditions until reaching the best before date is not efficient considering the required time and in commercial resources. Therefore, accelerated shelf-life testing (ASLT) has become a central step in the usual product development procedure. The objective of this work was the development and establishment of prediction models regarding the stability and shelf-life of orange juice and apple juice. To this end, the juices were stored at different temperatures and were investigated regarding their sensory profiles by quantitative descriptive analysis (QDA) and their compositions of volatiles by untargeted profiling via GC-MS. The final prediction models were derived by combination of the sensory and volatiles-related data sets in a holistic prediction approach. ukryj opis