In order to meet the needs of quality detection of agricultural products during logistics, a wireless electronic nose (WEN) based on GPRS for applications in the “on-the-go” environment was designed. It consisted of 5 modules which were the sensor-array with 8 MOS sensors, signal processing module, master controlling module, GPRS module and remote monitoring system. The GPRS communication performance of the WEN was tested, it was found that when data transmission cycle exceeded 400ms, three communication performance indexes of GPRS network which were the time-lag between sending and receiving, packet loss during the transit and transmission rate would meet the demands of detection simultaneously. An application aimed at fresh mangos was taken by using the present WEN, and the obtained profiles of mango aroma were clear and stable. Mango storage prediction models were built with stepwise regression method (SR) and LDA—SVM, respectively, and it was indicated that the SR model of the control group mango had a certain predictive effect only for the mangos stored for 4 and 6 days, the SR model of the experimental group and the LDA—SVM model either of the control group or the experimental group achieved satisfactory predictive effect as a whole.