Market segmentation is a diffuse topic, embracing a wide range of issues. This paper discusses a number of these that have arisen in work carried out by the author during recent years. She main theme of the paper is the need to preserve a flexible and eclectic approach in what are still early days of development. The basic rationale behind segmentation is the increasingly recognised point that consumers differ in ways that are exploitable by marketing and advertising strategy. It is thus important to analyse consumer data in terms of relevant criteria. These classifications should both increase our understanding of the market, and also be reproducible in other research enquiries. The main criteria used in segmentation studies have been behavioural, demographic, personality, and, most recently, attitudinal. The paper stresses the importance of taking an empirical approach to identifying target groups, and the need for using multiple criteria, to take account of the variety of factors influencing consumer decisions. The advantages and disadvantages are discussed of some of the main techniques used in segmentation. The methods of numerical taxonomy are particularly promising, as they take into account the interaction between different combinations of variables. Although the main focus of this paper is upon deriving target groups, the value of segmentation in terms of consumer perceptions of product differences is emphasised. So is the need to adopt a dynamic approach to segmentation, both by paying attention to the cyclical nature of the consumer decision process, and by examining the differential response rates of different market segments.