Before the 1980s, centralized forest policies in many African countries excluded local communities, while forest resources were frequently degraded. In response, Participatory Forest Management (PFM) was introduced to devolve management and improve livelihoods, forest condition and governance. Building on existing analyses that highlight the limited successes of PFM, my thesis explores the mismatch between policy objectives and outcomes in PFM. First, I conducted a policy review to explore whether, and how, devolution is specified in Tanzanian, Kenyan, Ugandan, Malawian and Ethiopian forest policies. The analysis considered the decentralisation framework developed by Agrawal and Ribot as well as the commitments of the Rio Declaration. In none of the five countries do the policies provide for all the critical elements required to achieve meaningful devolution, such as democratically elected, downwardly accountable local actors and equitable benefit sharing. I concluded that even without flaws in implementation, these decentralisation policies are unlikely to achieve true devolution in the study countries. I then used individual surveys, key informant interviews and focus group discussions to investigate whether Tanzanian PFM has achieved devolution in a specific case study, by comparing observed outcomes to stated policy objectives and the decentralization framework developed by Agrawal and Ribot. I found PFM had failed to achieve devolution, with institutions captured by a relatively centralised ‘elite-within-an-elite’, poorly accountable to ordinary residents and village authorities. Importantly, the failures were not just attributable to flaws in the policy identified by my review. I then assessed how Tanzania’s PFM policy was developed, to understand why and how the failings identified in policy design came about. I used key informant interviews with stakeholders that were involved in the process of PFM policy formulation. I found that foreign donors played a great part in driving the process of policy formulation, with a lack of wider stakeholder engagement. In addition, key personnel in the government actively disagreed with the rationale for PFM. This contributed to a weakened policy, with government reluctant to devolve appropriate powers to local communities. Finally, I used key informant interviews, focus group discussions and individual surveys to examine to what extent the process of PFM implementation at national, district and village levels contributes to PFM failure. I found that constraints on PFM implementation arose from a lack of capacity to support the approach, in terms of financial, human, and physical resources, as well as a lack of policy knowledge of local communities and forest staff. In addition, REDD+ policy (Reducing Emissions from Deforestation and forest Degradation) has also weakened the existing institutional framework for implementing PFM policy. Implementation of PFM has also been dependent on donor and NGO support, which has not been sustained. Overall, I found that PFM policy was weakened at each stage. A flawed policy making process, lack of political will and institutional resistance, led to formulation of policy that had weaknesses e.g. policy allowed unrepresentative local institution to develop in PFM as well as transferred unbalanced enforcement powers to local communities. Devolution was then stymied by existing power relations and resource constraints at each level of implementation, from national to local. Overall, while it was hoped that PFM would address deficiencies in centralised forest management by devolving power to local people, its success was limited by the very factors it aimed to address, e.g weak state capacity that aimed to address. The result may more closely resemble a privatisation of resources that were previously managed de facto as common property by surrounding communities than true devolution.