Publication and citation indicators of groups are thought to enhance the quality and legitimacy of science policy decisions. While these indicators might be of value from a policy pont of view, the relation between these cumulative data and the local circumstances that influence the development of scientific knowledge has not been explored extensively. In this paper it is argued that publication and citation patterns related to research units are influenced by local circumstances. Toxicology is chosen as an example because it is directed at solving social problems and relates to local practices. In this paper, output indicatiors of Dutch toxicological research units are related to qualitative information on the strategies of these units. it can be shown that the variation in output and citation indicators can be explained in terms of local variations in context. Such variations in local organizational settings should caution against the application of scientometric studies to measure impact as an indicator of scientific quality.