Molecular sensory science techniques such as quantitative sensory description， aroma extract dilution analysis technique， aroma activity values (OAVs) and aroma component recombination model were used to investigate the key contributing components of flavor differences between heat-treated and fresh bayberry juice by combining modern instruments such as gas chromatography-mass spectrometry (GC-MS)， gas chromatography-olfaction (GC-O) and electronic nose. The results of AEDA showed that after heat treatment， the FD of ‘green/grassy’ and ‘fresh’ aroma active components in bayberry juice decreased， and the FD of ‘sweet’ increased， and new aroma-active components such as ‘boiled melon’， ‘stimulating’， ‘putrid’ and ‘leather’ were produced. Combining the variable projection importance (VIP) and OAVs of the aroma active components， 13 key aroma variables were identified， including hexanal (green/grassy)， 3-hexenal (green/grassy)， (E)-2-octenal (sweet)， (E，Z)-3，6-nonadienal (watermelon peel)， (E)-2-nonenal (fatty)， (E，E)-2，6-nonadienal (cucumber/herbaceous)， (E，E)-2，4-nonadienal (shieldbug)， 2，5-dimethoxy-benzaldehyde (chicken fecal)， linalool (green apple)， 10，10-dimethyl-2，6-dimethylenebicyclo[7.2.0]undecan-5-ol (bitter)， pyrrolidine-2，4-dione (pungent/chemical)， β-damascenone (sweet)， (E)-2-(2-pentenyl) furan (melon) can be used as indicator compounds to distinguish heat-treated bayberry juice from fresh bayberry juice， with β-damascenone having the highest OAV value. This study could provide theoretical support for the product development and quality control of bayberry juice.