This case study compares Proshika's different experiences of interaction with government in two of its programmes -- Livestock and Social Forestry -- and then draws some general conclusions about the problems and potential for NGO-government relations in Bangladesh. Proshika works with over 14,000 organised groups of landless men and women and provides a 'package' of inputs and services to support both income generation activities and `conscientisation'. The Livestock Programme has provided the government with a well-networked distribution .system for its scarce inputs. This collaboration has served the interests of both government and NGO, allowing the government to distribute vaccines and services more widely, and strengthening Proshika's group-based activities by giving members proper access to inputs and support facilities through which they can generate income from livestock rearing. Most of the group members directly concerned with cattle rasing are rural women. The Social Forestry Programme has been a different experience. It has grown out of Proshika's long involvement in environmental projects, such as its tree plantation campaign, which began in 1976. It has several different areas of activity, such as tree plantation, agroforestry and roadside forestry. The aim of the Programme is for organised group members to gain access to available local public and state-owned resources, such as roadsides, ponds and state reserved forests. The struggle to gain secure access to the `sal' forests, which are managed by the Forest Department, has been a case of negotiation and confrontation by the group members. This contrasts with the livestock case. On the one hand, there has evolved a clash of interests between the group members and the local level Forest Department officials vis-à-vis their allies in the rural power structure. On the other hand, the group has entered into a new form of negotiation over the newly grown forest resources with members of the community in securing a more equitable distribution of intermediate benefits. The conflict of interests between the group members, the rural elites and the Forest Department can only be resolved on the part of the State. Similar experiences have been encountered in the Roadside Forestry Projects. As long. as the Social Forestry Programme involves the extension of services to the rural poor (eg. tree planting and training), the partnership between Proshika and the government runs relatively smoothly. However, Proshika's experience of working at the grassroots indicates that appropriate tenurial arrangements are an essential prequisite to a successful Social Forestry Programme. This paper argues therefore that while mutually beneficial links are possible between GOs and NGOs over input delivery, a restructuring of existing access relationships for poor people to Common Property Resources (CPRs) requires more challenging initiatives by government agencies.