Canister (6 l, Restek) and aluminum polyester bag (5 l, Top trading) sampling methods were compared to determine which sampling method is effective. Between the two methods, canister had very low background compared to bags (data not shown). However, it was difficult to collect odor sample by using canister since it has high sampling flow rate and the collected odor cannot be perceived by panelists to check if the odor has the same characteristics. Although a bag method had a little higher background level compared to the canister method, more odor peaks could be perceived from the used bags in GC/FID/O analysis. Furthermore, it was possible to collect the odor effectively using bags since the sampling flow rate is not so high and the collected odor could be perceived by panelists. The odor was collected in both canisters and bags at the same time (n = 3) and they were analyzed three times for each sample by GC/FID/O. No odor was perceived from canister samples. On the contrary, several odors were perceived from bag samples. The average perceived odor peaks were 3.7 times for the bag sample while canister sample had no odor peak. A bag blank was also analyzed by putting pure nitrogen gas in GC/FID/O, GC/MS, and GC/AED analysis. No odor was perceived in GC/FID/O and the level of bag blank was similar to SPME blank in GC/MS and GC/AED analysis. Furthermore, bag blanks by putting outside air and air from the laboratory room were analyzed in GC/FID/O analysis. The results from blank test in GC/FID/O are shown in Fig. 7. The black solid line chromatogram indicates FID signal and the red solid line at the bottom of each chromatogram indicates ODP signal.