The Mothusimpilo Project ("Working together for health") is designed to develop ways of responding to HIV/AIDS in a gold mining community of 250,000 people in Carletonville, South Africa. It uses participatory approaches to maximise active involvement of local groups and the community in programme design and implementation. The project employs three fulltime workers, recruited locally, and has two major components: improved STD prevention (through the coordination and upgrading of STD services (provided by government and mine hospitals and clinics, general practitioners and traditional healers) and communitybased condom distribution and peer education, with particular emphasis on migrant mine workers, commercial sex workers (CSWs) and youth. The project seeks to maximize community representation and participation at two levels: through stakeholder management of the programme as a whole and through grassroots involvement in particular aspects of programme implementation. Funding comes from local and overseas donors.

The first success has been in pulling together a management team of various stakeholders, including representatives of the provincial and national health departments, the gold mining industry, trade unions, academics, donor representatives and the community-based Carletonville AIDS Action Group. As a result, the project now has the commitment of all the major stakeholders, access to a range of local and international expertise and strong community involvement at the grassroots level. However, this process took almost two years to complete and the experience is instructive. In 1993, Yodwa Mzaidume, then a nursing sister working for AngloAmerican, the biggest gold mining company in South Africa, saw the threat HIV/AIDS posed to Carletonville and decided to try to deal with the problems. Mzaidume was born in Carletonville. Her father worked on the mines there and she had good contacts and a strong identification with the community. The Carletonville AIDS Action Group (CAAC), which grew out of her efforts, included community groups ranging from organisations representing nurses, traditional healers, teachers and social workers, and trade unions, political parties, women's groups, various religious groups and local government representatives.

Despite repeated attempts from 1993-95 by CAAC to get financial support from the National and Provincial Departments of Health respectively, possibly because the format and budgeting of their proposal could have been improved on. They were unable to access the expertise that was needed to make the proposal convincing.

Even having achieved this level of collaboration and cooperation, the management of a multistakeholder project is a complex process. Stakeholders differ in their views about the nature of the problem and how to address it, their motives for being involved, and the commitments they are willing to make. Some see HIV/AIDS as a biomedical problem to be dealt with through drugs and clinics, others as a problem of poverty and underdevelopment, to be dealt with through the empowerment of people in highrisk situations. Project workers have no authority over stakeholders whose involvement is voluntary, and they often lack the status and confidence to elicit the levels of stakeholder co-operation which would optimise the likelihood of project success. To help understand these complexities there is an ongoing process evaluation of the project, which includes an annual stakeholder/management analysis by an independent evaluator.

What is the context of this work? Migrant mine workers travel long distances from their rural homes to work on the mines in Carletonville where they are housed in single sex hostels. Impoverished women flock to the mines to make a living by selling sex and alcohol to men. The men are reluctant to use condoms and the project has made good progress in working with sex workers to promote their use. A case study of the communityled condom promotion is being conducted amongst sex workers in the area, many of whom live in squatter camps in conditions of extreme poverty and violence. These shack settlements are often controlled by unofficial groups of selfelected men, who act as local community "gatekeepers". These men wield a fair degree of control over community residents, often through the use of violence. Access to such communities by project workers has involved intensive periods of complex negotiation between project workers and gatekeepers, who are often hostile to any community activities they do not directly control themselves. In such a strongly maledominated community, such men do not always welcome the notion of a health project controlled by women, particularly women who are also sex workers. After many hours of careful and sensitive negotiation, the programme has been positively received by the men; 38 sex workers have being recruited and trained to implement the programme in various shack areas. They work through largerscale community meetings, and onetoone individual contacts.

While the project has been designed to maximize stakeholder involvement in project management, and grassroots control of local communitybased peer education programmes, enormous obstacles remain. The project is only in its second year and efforts are being made to understand, reflect on and address these obstacles. In addition to addressing the problem of HIVtransmission, the project should generate lessons on factors that promote or hinder community involvement as the backbone of any sexual health and development initiative.

Brian Williams, Mothusimpilo project, CSIR, P. O. Box 91230, Auckland Park 2006, Johannesburg, South Africa; Tel: +27-11-358.0149; Fax: +27-11-482.3267 or 726.7308  e-mail: bgwillia@csir.co.za and Catherine Campbell, Social Psychology, London School of Economics, Houghton Street, London WC2A 2AE, UK; Tel: +44-171-955.7701; Fax: +44-171-955.7565; e-mail: c.campbell@lse.ac.uk