Constant market shares (CMS) analysis compares the actual export growth performance of a country with the performance that would have been achieved if the country had maintained its exports relative to some standard. The approach was first applied to international trade in the 1950s and has generally been used to analyse trading patterns and, in particular, the extent to which poor export performance can be attributed to a loss of ‘competitiveness’. However, the approach has been open to objections as a tool of description and diagnosis. Recent revisions appear to meet objections concerning its role as a descriptive tool. However, CMS analysis remains open to objections as a diagnostic tool owing to the strict theoretical conditions required to yield an unambiguous interpretation. In this paper we generalise the constant market share framework based on recent revisions by Jepma (1986). Alternative models are derived and interpreted with particular attention to the underlying theoretical conditions required for diagnostic interpretation. We conclude that the prospects for CMS analysis as a diagnostic tool depend upon further research into its theoretical foundations, the extent to which the implicit aggregation assumptions can be tested and, of immediate concern, development of a computer program. Keywords: CMS analysis, trade, aggregation, Armington model.