The purpose of this study was to investigate the serum metabolic difference between hepatocellular carcinoma (HCC, n = 20) male patients and normal male subjects (n = 20). Serum metabolome was detected through chemical derivatization followed by gas chromatography/mass spectrometry (GC/MS). The acquired GC/MS data was analyzed by stepwise discriminant analysis (SDA) and support vector machine (SVM). The metabolites including butanoic acid, ethanimidic acid, glycerol, L-isoleucine, L-valine, aminomalonic acid, D-erythrose, hexadecanoic acid, octadecanoic acid, and 9,12-octadecadienoic acid in combination with each other gave the strongest segregation between the two groups. By applying these variables, our method provided a diagnostic model that could well discriminate between HCC patients and normal subjects. More importantly, the error count estimate for each group was 0%. The total classifying accuracy of the discriminant function tested by SVM 20-fold cross validation was 75%. This technique is different from traditional ones and appears to be a useful tool in the area of HCC diagnosis.