Abstract: To improve the precision and robustness of electronic nose, six-point smoothing method was used to de-noise the gas sensors data, and support vector machine(SVM) was used to develop recognition models. Compared with those original gas sensors curves, the curves after pretreatment were smoother, but their shapes showed not much difference. This indicated that the major information in apple could be reserved while noise was removed by the six-point smoothing method. The maximum of each sensor response was extracted as features. Principal component analysis(PCA) and two support vector machine(SVM) models were used to analyze the features and distinguish three different cultivar apples which were "Fuji", "Huaniu" and "Jina". The PCA results show that it is difficult to distinguish the three apple cultivars by linear models. The recognition of "Haniu" was 100% obtained by the first SVM model, and the distinguishing ability of second SVM model between "Jina" and "Fuji" was 90% for calibration set and 100% for prediction set.