The fluorescence characteristics of natural organic matter in a groundwater based drinking water supply plant were studied with the aim of applying it as a technique to identify contamination of the water supply. Excitation–emission matrices were measured and modeled using parallel factor analysis (PARAFAC) and used to identify which wavelengths provide the optimal signal for monitoring contamination events. The fluorescence was characterized by four components: three humic-like and one amino acid-like. The results revealed that the relative amounts of two of the humic-like components were very stable within the supply plant and distribution net and changed in a predictable fashion depending on which wells were supplying the water. A third humic-like component and an amino acid-like component did not differ between wells. Laboratory contamination experiments with wastewater revealed that combined they could be used as an indicator of microbial contamination. Their fluorescence spectra did not overlap with the other components and therefore the raw broadband fluorescence at the wavelengths specific to their fluorescence could be used to detect contamination. Contamination could be detected at levels equivalent to the addition of 60 μg C/L in drinking water with a TOC concentration of 3.3 mg C/L. The results of this study suggest that these types of drinking water systems, which are vulnerable to microbial contamination due to the lack of disinfectant treatment, can be easily monitored using online organic matter fluorescence as an early warning system to prompt further intensive sampling and appropriate corrective measures.