Grafting is widely applied in the cultivation of melon. In this study, ‘Qinmi No.1’ (Cucumis melo L.(QG)) and ‘Ribenxuesong’ (Cucurbita maxima Duch. (RG)) were used as rootstocks for ‘Qingxin Yangjiaocui’ (Cucumis melo L.). The results showed that grafting with muskmelon rootstocks had no significant effect on fruit aroma, but grafting with pumpkin rootstocks significantly reduced the odor intensity and odor preference scores of melon fruits. Compared with the fruits from self-grafted plants (SG), four new aromatic volatiles with a sweet smell were detected, the alcohol dehydrogenase (ADH) activity was significantly decreased at 30 DAP, but unaffected at 42 DAP in QG fruits. There was no difference for alcohol acetyltransferase (AAT) activity between QG and SG fruits. The expression level of CmADH2 was significantly higher at 30 DAP and 42 DAP, but CmAAT2 was significantly lower at 42 DAP in QG fruits compared with SG fruits. In RG fruits, the main aroma compounds including butanoic acid ethyl ester, 2-methyl-2-butene-1-al, and 2-methylheptan-1-al were absent, while the volatile compounds with unpleasant odor characteristics including trans, cis-2,6-nonadien-1-ol, (E,E)-2,4-heptadienal, octanoic acid, and styrene were detected. Compared with SG fruits, 1-nonanol and 1-heptanol with green odor characteristics were significantly increased, but eucalyptol and farnesene with fruity aroma characteristics were significantly decreased in RG fruits. The ADH activity of RG fruits was significantly lower than that of SG fruits at 30 DAP and the AAT activity was significantly lower than that of SG fruits at 42 DAP. In addition, the expression levels of CmADH and CmAAT homologs in RG fruits were significantly lower than those in SG or QG fruits. These results show that grafting with pumpkin rootstocks affected the main aroma components, reduced ADH and AAT activities, and down-regulated the expression levels of CmADHs and CmAATs in the melon fruits. This study reveals the mechanism of different rootstocks on melon fruit aroma quality, and lays a theoretical foundation for the selection of rootstocks in melon production. Future studies using overexpression or CRISPR/CAS system to obtain stable transgenic lines of genes encoding key aromatic volatiles, would be promising to effectively improve the flavor quality of melon.