This article summarises some of the advances made over the past few years in understanding glass compositions and their application to the investigation of trade and exchange. It focuses on Roman glass, which is understood here to include not only the glass produced during the Roman period in the strict sense but also glass of similar composition that was in use throughout most of the first millennium AD and up to the mid-ninth century. The author presents recent developments in the understanding of these low-magnesia soda-lime-silica or natron glasses. It appears that glass production was divided between a small number of primary glass making centres, situated mainly in the Near East, and a large number of secondary fabrication workshops that re-melted and shaped the lumps of raw, pre-melted glass. Glass may be related to its primary production group by elemental analysis and, where there are data from workshops, to the production centre or region. The recycling of old glass is revealed by trace element analysis due to the contamination in primary glass compositions by small quantities of coloured glass incorporated in the recycled material. The isotope analysis of strontium (Sr) and lead (Pb) allows the geological environment of the raw material to be inferred and in some cases, provenance to be predicted. -- ICCROM