As a novel type of industrial establishment in the era of globalization, foreign direct investment (FDI) has become a new driving force in the shaping of urban structure. To understand the current worldwide urban restructuring process, the location behaviour of FDI ®rms should be fully investigated. But few empirical studies on industrial location have been carried out at the intrametropolitan scale, probably due to the lack of disaggregated data and appropriate models. This study geo-references FDI ®rms by their postal codes and captures site attributes through cross-referencing ®rm distribution with other spatial coverage such as population statistics, land uses, and designated development zones. This paper argues for a Poisson and negative binomial model of ®rm location based on count-data and tests such a ®rm location model using the case of metropolitan Guangzhou, PRC. Results, in contrast to previous studies at the national and regional scales of this country, con®rm that, at least statistically, intra-urban FDI ®rm location in the context of a transitional economy presents regularities that are explainable according to`rational' economic considerations of site suitability.