Chinese national food safety standard for raw milk regulates that the milk is regarded as hygiene disqualified milk if its total bacterial count (TBC) exceeds 2 x 10(6) CFU/ml. To provide a rapid method for identifying hygiene disqualified milk, the dielectric spectra of 150 raw goat's milk samples were obtained. The partial least squares discriminant analysis, support vector machine (SVM), and extreme learning machine algorithms were applied to build models to identify whether the milk was hygiene qualified or disqualified on TBC. The results showed that SVM based on principal component analysis was the best model with total identification accuracy rate of 100%. The research indicates that the dielectric spectra could be used to detect whether the TBC of milk exceeds the national standard or not, and provides useful information on developing a rapid detector to evaluate milk hygienic quality in-situ or on-line.