The effects of beaker volume, sample weight and headspace-generation time on the response of a PEN2 electronic nose (e-nose) were studied. Both the peak and final stable responding points of sensors were extracted. Principle component analysis (PCA), linear discriminant analysis (LDA) and analysis of variance (ANOVA) were employed to analyze the responding of each sensor as well as the whole sensor array, and the optimum parameters (500mL beaker volume, 25g sample mass, 5min headspace-generation time) were acquired for later beef storage time detection. The difference in volatile profiles of beef samples of seven different storage times (0, 3, 5, 7, 10, 12 and 14d) was confirmed. The stepwise linear discriminant analysis (Step-LDA) results showed that the correct rate was 98.57%.