This dissertation critically reviews three community safety partnerships. It contributes to learning by using new data to examine some of the enduring challenges faced by the partnerships as they respond to the changing socio-political context. These multi-agency bodies primarily involve the police and local authority, along with the fire service, and primary care trust as statutory partners, and other criminal justice agencies. Although multi-agency work has a long history, Community Safety Partnerships originated most notably within the Morgan Report published in 1991. The principles outlined within the Morgan Report were subsequently embodied in the Crime and Disorder Act 1998 which itself attempted to formalise and standardise community safety structures and practice. The research has involved fifty-eight in-depth interviews with key stakeholders, and observation of twenty-nine meetings held in Birmingham, Cambridge and Lincoln. These three areas were chosen to facilitate comparison in terms of the size of the city, local authority structure, level of urbanisation and concomitant social deprivation, and levels of crime. My research suggests four key challenges. The first of these concerns the development of structures within the post Crime and Disorder Act 1998 period. The purpose, structure and processes varied between areas. Birmingham, for example, struggled to develop an appropriate structure because of its size and the devolution of local authority services. In all three areas, however, those interviewed noted a lack of decisionmaking and implementation which raised questions about the purpose of the partnerships beyond being 'talking shops' . The second challenge concerns the changing social context for new partnership developments. For example, the two-tier local authority structure in Cambridge and Lincoln posed particular problems. Moreover, in all three areas community involvement appeared to be symbolic rather than 'real'; this inhibited developments and emphasised some of the difficulties inherent in communitariarusm. The third challenge relates to funding and performance monitoring arrangements. Here, practitioners noted the influence of bureaucracy and 'short-termism'. The early 'honeymoon period' where there was relatively little government interference (Phillips et al., 2002) had ended and the partnerships had clearly experienced increasing managerialist pressure, but in spite of this pressure, evidence of longer-term success remained scarce. As outlined in the Audit Commission (2002), practitioners in the three partnerships acknowledged that with the exception of specific initiatives, the post 1998 developments had yet to make a significant impact on crime and disorder or that at best, they remained unclear about the impact. Such uncertainty about impact could be a consequence of the difficulties of measuring performance, of course, due to difficulties in accessing relevant data and information about community safety initiatives. Fourth, there appear to be inherent difficulties in assuming that 'many agencies are better than one' in addressing community safety (Liddle, 2001). An 'ideology of unity' (Crawford and Jones, 1995), however, may mask underlying tensions. My research revealed tensions at different levels, including tensions between the local partnerships and national government. This is not to say that local practitioners lacked autonomy, however, as they were able to resist some of the governmental constraints. But interagency relationships appeared to be underpinned by power struggles which served to undermine joined-up community safety practice; in particular, the struggles raised questions about who was responsible for community safety in each area. The challenges for the partnerships, as revealed in this dissertation, suggest that the recommendations within the Morgan Report of 1991 have not been addressed nor has the Crime and Disorder Act 1998 standardised community safety structures and practice. The notion of 'responsibilisation' (Garland, 2001) through decentralised governance is clearly a complex issue; the Government appears to wish to both 'steer' and 'row' each of the partnerships and this leaves practitioners uncertain of their own role. This is one example of the contradiction between the 'reality' and symbolism of community safety practice which seems to underpin the partnerships.