Abstract Plant-mediated indirect interactions (PMIIs) between phytopathogenic fungi and herbivorous insects on shared host plants occur in nature. Knowledge of PMIIs is critical in plant molecular breeding and integrated pest management. We studied the response and chemical mechanism of beet armyworm, Spodoptera exigua (Hübner), adults to rose plants, Rosa chinensis Jacquin, infected with rose powdery mildew, Podosphaera pannosa (Wallr.: Fr.) de Bary. Using gas chromatography-electroantennographic detection (GC-EAD) coupled with electroantennogram (EAG), we found that beet armyworm antennae responded to 8, 11, and 3 volatile organic compounds (VOCs) from noninfected roses, mildew-infected roses, and mildew alone, respectively. The EAG analyses showed 11 chemicals (e.g., limonene [1], 2-ethyl-1-hexanol [2], linalool [3], nonanal [4], (E)-β-caryophyllene [5], 1-dodecanol [7], nhexadecane [9], 1-hexadecanol [11], methyl palmitate [12], 1-octadecanol [14], and n-butyl hexadecanoate [15]) elicited electrophysiological responses of beet armyworm antennae with significant dose-response relationships (P < 0.05). The EAG responses to the three chemicals (3, 11, and 15) were greater than that to the reference chemical [i.e., (E)-2-hexenal] at 0.5, 5.0, and 50.0 mg/ml. Olfactory and ovipositional behavior assays indicated that three chemicals (2, 3, and 5) significantly attracted beet armyworm females and four chemicals (7, 11, 14, and 15) strongly repelled females. Chemicals 2, 3, and 5 from healthy roses appear to be responsible for the attraction of beet armyworm moths to healthy roses, whereas chemicals 7, 11, 14, and 15 from mildew-infected roses play key roles in inhibiting attraction of moths. VOCs from mildew alone did not attract or repel beet armyworm moths.