Glass is one of the most common types of trace evidence found at crime scenes and on suspects. Refractive index is presently used for matching recovered and control samples of glass. However, increased quality control in manufacture has substantially reduced the spread of RI values and many samples can no longer be distinguished. Therefore analysis of glass by elemental techniques is required if samples are to be distinguished. This research project was carried out on behalf of the Institute of Environmental Science and Research (ESR), Auckland. The aim was to develop a method to analyse automotive glass fragments by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS), and create a database which can be used to categorise samples into country of origin and to distinguish samples. 244 glass samples were obtained from ESR which contained 162 laminated and 82 toughened samples from a range of vehicle makes and models manufactured between 2002 and 2006. Two calibration standards (NIST 612 and FGS 2) were compared for their ability to calibrate float glass. FGS 2 is a better standard due to its closer composition to float glass and was used for this research. 23 elements were initially analysed and assessed for accuracy, precision and long term stability. This was cut down to 15 elements for the final method. The relative accuracy and precision of these elements was less than 5 % when compared to the literature values. An inter-laboratory comparison study showed most elements had less than 10 % bias between laboratories when analysing float glass. However, Sn had over 25 % bias between the two laboratories. A homogeneity study of three panes showed no spatial variation for most elements. However, the three Pb isotopes showed spatial variation in the windscreen. In addition, outliers were found for replicates of some database samples and the pooled standard deviation for the database was very large, indicating Pb is not homogeneous in glass. Multivariate analysis was used to investigate the natural splitting of the data. The splitting was influenced by the country of origin from which the raw materials were sourced from for glass manufacture. Australian samples had a clear separation from all samples originating from the Northern Hemisphere. A multiclass classifier correctly categorised 86.87 % of samples into the vehicles‟ country of origin. Correct classification was not higher due to importation of some glass. For example, Australia imports glass from Thailand for the Falcon and Commodore ranges. Statistical methods were compared in their ability to discriminate fragments. The elemental composition range overlap method allowed 80.39 % (164/204) discrimination between samples. Only small elemental variation between Australian samples was observed. With Australian data removed, 93.59 % (146/156) of samples could be distinguished. A three-step method was created to increase the distinguishing power; this used a combination of elemental composition and RI. This procedure was able to distinguish 84.31 % (172/204) samples from the entire database and 94.87 % (148/156) samples with Australian data removed. This method decreased the occurrence of type II errors.