Gasoline is a substance commonly encountered in forensic settings. Unfortunately, gasoline is an easily obtainable ignitable liquid that arsonists commonly use to initiate or expedite the spread of an intentionally set fire. Fires claim the lives of many people each year in addition to causing widespread property damage. Many fire scene investigations result in charges of arson, which has the legal connotation of a committed crime. For this reason, extensive analysis and investigation must be undertaken before any suspected arson scene is deemed an actual case of arson. Although ignitable liquids, including gasoline, may be present at the scene of a fire, it does not necessarily mean they were intentionally used as accelerants. An accelerant is a fuel used to initiate a fire. These realities, in addition to several other factors, demonstrate why a rapid, reliable, gasoline analysis method is crucial to forensic applications. In this thesis, direct analysis in real time – mass spectrometry (DART-MS) is evaluated as a potential method that could better identify, distinguish and classify gasoline brands from one another. Techniques such as DART-MS could enable forensic laboratories to better identify questioned gasoline samples. Many ignitable liquids share similar chemical properties, and forensically relevant evidence is often obtained from a crime scene in less than favorable conditions. Fire debris can encompass various materials, including burnt carpet, flooring, items of furniture and clothing, among others. If gasoline was used as an accelerant, it may be present in trace amounts after the termination of the fire. Materials submitted for laboratory analysis may be substrates with compositions that have components similar to those found in some ignitable liquids. These are just a few of the potential obstacles that could be encountered with analyzing fire debris in a forensic setting. Traditionally, gas chromatography – mass spectrometry (GC-MS) methods are utilized for gasoline analysis in the criminal laboratory setting. While traditional GC-MS methods are sensitive and able to classify samples as gasoline, they are time consuming in terms of both sample preparation and analysis. Additionally, they do not generate differential mass spectral data based on the brand of gasoline. Conversely, gasoline analysis in this research, utilizing the DART-MS method, demonstrated that five different brands of gasoline could be distinguished from one another both by visual examination of mass spectra and with methods of chemometric analysis. Advantageously, the DART-MS method, an ambient ionization technique, requires little sample preparation and a rapid sample analysis time, which could drastically increase the throughput of standard sample analysis with further method development. The goals and objectives of this research were to optimize the DART-MS parameters for gasoline analysis, determine if DART-MS analysis could distinguish gasoline by brand, develop chemometric models to appropriately classify gasoline samples, and finally lay groundwork for future studies that could further develop a more efficient and discriminating DART-MS gasoline analysis method for forensic casework. Each brand of gasoline was observed to have a chemical attribute signature (CAS) consisting of not only low-mass ions, but also a variety of high-mass ions not usually observed with gasoline samples analyzed by GC-MS. Although variables including season, storage time, dilution and age of the gasoline were observed to contribute to the resulting mass spectral data, once the mass spectra are better understood, they could offer even more discriminating power between samples than simple analysis of the gasoline brand. In this research, DART-MS parameters were first optimized for gasoline analysis. Subsequently, the five acquired brands of gasoline: Shell, Sunoco, Irving, Cumberland Farms and Gulf, were analyzed both undiluted (or neat) and diluted utilizing the DART-MS analysis method. GC-MS data was generated and analyzed to show comparisons. After analyzing the data generated by both approaches, it was apparent that the DART-MS method could generate CASs based on the gasoline brand and offer a degree of differentiation that traditional GC-MS does not. Additional chemometric analyses utilizing principle component analysis (PCA) and the construction of models with Analyze IQ Lab software verified that the gasoline brands were distinguishable when samples were analyzed with this ambient ionization method. PCA plots of the neat gasoline demonstrated clustering based on brand. Additionally, models constructed from training samples generated from DART-MS analysis of the various brands were able to accurately classify gasoline samples as "yes" or "no" when a test set of gasoline was compared to all five brands. The lowest associated testing error rate for some of these models was 0%. However, additional analysis with greater sample sizes needs to be further carried out to more accurately evaluate this method of gasoline analysis and classification.