Abstract: The aroma of rose water is positively correlated with its quality and economic benefits, and it is the main basis for determining the quality of rose water. The volatile components in rose water were analyzed by headspace solid phase microextraction-gas chromatography-mass spectrometry (HS-SPME-GC-MS) combined with automatic mass spectrometry deconvolution and identification system (AMDIS). Taking Kushui rose flower water as the research object, the effects of different solid phase extraction heads, sample volume, magnetic stirring method, sodium chloride content, extraction temperature, equilibrium time, and extraction time on the extraction effect of rose flower dew aroma components were optimized. Peak area and peak number were two indicators to investigate the extraction effect of each factor. The chromatographic peaks were purified by AMDIS under the optimal experimental parameters: solid phase extraction head of 75 μm Carboxen/PDMS, sample size of 10 mL, magnetic stirring, sodium chloride content of 1.0 g, equilibrium time of 20 min, extraction temperature of 40 ℃, extraction time of 40 min. 60 kinds of aroma components in rose flower water and rose cell liquid were identified with the Spectral Library Search and retention index. 59 kinds of rose flower water were identified, and the relatively high content of aroma components were citronellol (19.00%), methyl eugenol (13.97%), eugenol (13.81%), α-terpineol (11.80%), linalool (5.27%). 22 kinds of rose cell sap were identified, and the relatively high content of aroma components were citronellol (44.22%), benzyl alcohol(6.07%), methyl eugenol (5.08%), α-terpineol (5.08%), (E)-linalool oxide (3.39%), linalool (3.10%). There were 21 kinds of components in rose water and rose cell fluid, the sum of relative content was 76.60% and 78.42%, respectively. The results showed that the identification results obtained by automatic deconvolution technology, library search and retention index were mutually verified, which could significantly improve the accuracy and reliability of the identification of unknown components, and make up for the co-elution of impurities, noise and compounds in low-resolution mass spectrometry. In the qualitative process of the target, there were multiple compounds with high matching degree in the same mass spectrum, or the phenomenon of similar matching and extremely low matching occured when the target was qualitatively identified, which made accurate qualitative determination of the target difficult, and even leaded to qualitative errors. This qualitative method can be widely used in the qualitative analysis of volatile components in complex samples.