Heavy metal pollution in water seriously affects human health. The disadvantages of traditional metal ion detection methods involve long and cumbersome chemical pretreatment in the early stage, and large volume of samples. In this study, microalgae were used as the medium, and terahertz spectroscopy technology was employed to collect the changes of material components in it, so as to deduce the types and concentrations of heavy metal pollution in water. Through the partial least square(PLS), we establish the prediction model of heavy metal concentration, and the results show that the best detection time for Pb is 6 h and Ni is 18 h. The principal component analysis(PCA) shows that β-carotene is the most affected substance. Afterward we collect five real surface waters in East China and verify that the judgment accuracy of Pb and Ni are 100% and 93.2% respectively. The results indicate that the time is shorter than the traditional pretreatment time from more than 20-6 h, the sample volume is reduced from 50 mL to 10 mL, the detection accuracy is improved from 10 ng/mL to 1 ng/mL. In a word, we provide a new fast and real-time method for biological monitoring of heavy metal pollution in water.