The assessment of fish microbial spoilage is a major fish industry interest to minimize the loss and increase the productivity of the catch. The existing spoilage assessment methods are time and labour-consuming. Thus, developing a rapid method to detect fish spoilage remains an essential requirement of the fish industry to minimize fish catch loss and facilitate its marketing. This study aims to evaluate the efficiency and reliability of the electronic nose (E-nose) to detect fish microbial spoilage in comparison with the conventional methods using fresh sardines (Sardinella longiceps). The efficiency and reliability of E-nose were evaluated by comparison with three parameters; total bacterial count (TBC), hydrogen sulfide (H2S) producing bacterial count and trimethylamine oxide (TMAO) reducing bacterial count in six storage studies (i.e. six different lots) divided into two temperatures (0oC and 25oC). Total bacteria were enumerated on tryptone soya agar, whereas iron agar and TMAO agar were used to enumerate H2S producing bacteria and TMAO reducing bacteria respectively. Principle component analysis (PCA), linear discriminant analysis (LDA) and artificial neural network (ANN) were used to demonstrate the E-nose data. Significant changes in TBC and H2S producing bacterial count occurred between 4 and 6 days of sardines stored at 0oC. However, significant changes of TMAO reducing bacterial count were found between 2 and 4 days of storage. On the other hand, changes in TBC and TMAO reducing bacterial count were significant between 3 and 6 h of storage at 25oC, while the count of H2S producing bacteria showed significant changes between 6 and 9 h of storage. At the end of storage at 0oC, the count of TBC, TMAO reducing bacterial count and H2S producing bacterial counts were 6.25, 5.5 and 5.6 log CFU/g respectively. In contrast, the counts were 5.79, 4.98 and 4.80 log CFU/g for TBC, TMAO reducing bacteria and H2S producing bacteria respectively at the end of storage at 25oC. The LDA and ANN showed a good classification of E-nose data for different storage times at two storage temperatures (0oC and 25oC). However, PCA classified was less efficient than LDA and ANN in spoilage of fresh sardines. Correlation coefficients between E-nose and TBC, TMAO reducing bacterial count and H2S producing bacterial count at 0oC were 0.919, 0.960 and 0.915 respectively. However, at 25oC, the correlation coefficients were 0.859, 0.945 and 0.849. In conclusion, a high positive correlation was found between E-nose and all microbial groups which qualifies the E-nose as an efficient device in the detection of fresh sardine's microbial spoilage at different storage temperatures.