This paper discusses the relevance of brand image measurement to the current marketing role, and the importance of perfecting suitable measures and analysis procedures for different markets. It is argued that many of the conventional approaches can be improved; in particular, by concentrating more on analysis of individual data sets, together with linking imagery to preference information. Following a review of mainstream current approaches, an account is given of relatively new analytic procedures - such as the PILOT and LOCATOR models - which adopt micro-modelling principles in the treatment of classical brand image data to generate a linkage between imagery and preference. A case example is given to support this. Finally, the paper discusses some of the experience gained over the last decade, and makes some suggestions for the future.