Caraway is one of the most consumed medicinal spice around the world. Different cultivars of caraway represent differences in the aromatic profiles which mark the difference in their selling prices in the international market. The present study provides a non-destructive methodology based on metal oxide semiconductor (MOS) based electronic nose (e-nose) and chemometrics for rapid classification of five different caraway cultivars (Zarand, Jupar, Kerman, Khabarbaft and Kohbanan) from Iran. To acquire enhanced aromatic profiles the study utilises temperature modulation technique in a MOS based electronic nose. The temperature modulations were performed with two voltage functions i.e. rectangular and sinusoid, and the best voltage function for classification was evaluated and reported. Measurements were performed in three phases and recorded transient responses of eight MOS sensors were considered for data analysis. To visualise the MOS sensor array data principal component analysis (PCA) was used. For classification purpose, linear discriminant analysis (LDA) and multi-class support vector machine (SVM) were performed on the PCA transformed data. Results showed that data visualisation with PCA clearly identify separate clusters over the PC transformed space. A maximum classification accuracy of 97.92 ± 3.82% was obtained for the SVM classification model on the rectangular voltage function. The performance of LDA modelling was similar to the SVM results in the case of rectangular voltage function, but in the case of the sinusoid form, the SVM results showed better accuracy to classify the caraway cultivars. As a consequence, it can be concluded that MOS based e-nose and chemometrics could support the rapid and non-destructive classification of caraway cultivars. Furthermore, the temperature modulation with rectangular voltage function can provide higher classification accuracy.