The article investigates the location behaviour of firms in the automotive industries in Thailand. Our approach is to use a logit model in order to analyse how the characteristics of the firms and the regions are related to the location decisions of firms in these sectors. The logit results throw some light on the question of the nature of agglomeration behaviour in a developing economy where the national spatial industrial structure is dominated by a single primal city. Our conclusions provide tentative support for a product-cycle argument of industrial concentration and dispersion. 1. Introduction During the 1980s and first half of the 1990s, Thailand has been one of the fastest growing economies in the world (Phongpaichit and Baker 1996). This has been accounted for mainly by the emergence of a large manufacturing (Tambunlertchai 1993) and service sector (Phongpaichit and Chiasakul 1993) base. Although the economy recently suffered a severe recession following a currency devaluation, the previous two decades of growth resulted in an enormous increase in Thailand's industrial development. In spatial terms, this growth had been focused primarily on the metropolitan area of Bangkok and its immediate regions. As the industrial sectors of Thailand began to modernize from the 1960s onwards, the geographical focus of investment, and in particular of foreign investment in the form of joint ventures, was in the area of Bangkok. This provided both the largest domestic market and the point of access for imports and exports. This process of spatially focused investment growth has continued during the more liberalized era of the 1980s and 1990s (Chalamwong 1992), and one consequence of this has been a large migration of labour from the predominantly rural outlying regions (McGee and Greenberg 1992; Krongkaew 1995b). By the end of the 1980s the Bangkok Metropolitan Region (BMR) accounted for 10 per cent of the Thai population, 49 per cent of Thai gross domestic product (GDP), 78 per cent of Thai manufacturing, 66 per cent of financial services and 52 per cent of other services, while average incomes in Bangkok were twice those of the national average (Rigg 1991, p. 157). The result of this high level of regional industrial concentration is that Thailand's distribution of activity is very uneven. Depending on the spatial definition employed,' Bangkok is between twenty-three (Rigg 1991, p. 138) and forty-five (McGee and Greenberg 1992) times larger than the second-order urban area, which is either Nakon Ratchasima or Chiang Mai depending on the definitions used (Kaothien and Webster 1994). This results in one of the most skewed national urban hierarchies of any country2 and some of the world's worst problems of urban congestion (Pianuan et al. 1994; Kidokoro 1992; Hayashi 1996). At the same time as Bangkok has been growing over the last decade, evidence suggests that there has also been something of a gradual decentralization of some types of activities away from the BMR region (Kaothien and Webster 1994; Poapongsakorn 1995; Puntasen and Riefler 1997). This tendency towards decentralization into some of the neighbouring regions is dominated by the manufacturing industry. It is the result of a gradual process in which firms of varying characteristics make a variety of different locational investment decisions over time, depending on the prevailing economic conditions. We investigate this regional industrial restructuring process from a microeconomic standpoint. Our approach involves accounting for the regional locational decisions of individual Thai-based manufacturing firms, in terms of the characteristics both of the regions and the firms themselves at the time when the locational decision was made. By aggregating these individual location decisions over time, it is possible to indicate some of the key factors which have encouraged this process of regional industrial dispersion for different kinds of firms within these sectors. …