There is an imperative need for a chemical sensor capable of remote, in situ, long-term monitoring of chemical species at sites containing toxic chemical spills, specifically at chemical waste dumps, landfills, and locations with underground storage tanks. In the current research, a series of experiments were conducted measuring the near-infrared optical absorption of alkanes, aromatics, and chlorinated hydrocarbons. A spectral library was then developed to characterize the optical spectra of liquid hydrocarbons. Near-infrared analysis was chosen due to compatibility with optical fibers. The goal was to differentiate between classes of hydrocarbons and to also discriminate between compounds within a class of similar molecular structures. It was observed that unique absorption spectra can be obtained for each hydrocarbon, and this uniqueness can be used to discriminate between hydrocarbons from different families. Statistical analyses, namely, principal component analysis (PCA) and correlation coefficient (Spearman and Pearson methods), were attempted to match absorption spectra from an unknown hydrocarbon with the database with limited success. An algorithm was subsequently written to identify the characteristic peaks of each hydrocarbon that could be used to match data from an unknown chemical species with the database.