Groundwater, a vital freshwater resource, faces increasing contamination risks from chemical industrial parks discharging hazardous compounds such as phenol and toluene. Detecting these pollutants at low concentrations is essential to ensure water quality and protect against long-term hazards. A method combining fluorescence spectroscopy and Gaussian feature extraction is proposed for the identification and quantification of phenol and toluene in groundwater. Fluorescence excitation-emission matrix (EEM) spectra of phenol and toluene are first measured, followed by feature extraction using a Gaussian function. The extracted features are then employed for qualitative identification and quantitative determination via support vector machine (SVM) and partial least squares (PLS) regression, respectively. For qualitative identification, Gaussian feature extraction is compared with original feature and PCA-based feature extraction methods. For quantification, it is compared with peak picking and PCA-based feature extraction methods. The results show that after Gaussian feature extraction, the performance is significantly improved. The identification accuracy for single-component samples reached 95.24 %, while for mixture samples, the accuracy was 90 %. In quantitative analysis of mixture samples, the average relative error for phenol concentrations of 2 µg/L or higher and toluene concentrations of 600 µg/L was controlled around 10 %, while for phenol concentrations at 1 µg/L, the relative error was about 30 %. This approach enhances both identification and quantification performance, providing a reliable tool for the early detection and quantification of low-concentration contaminants in groundwater, with great potential for environmental protection.