It has become increasingly common to use projects as a form of organization when implementing public policies in the area of social welfare and probably also in other policy areas. A key player behind this development is the EU and its various structural funds, but the same trend can also be found at national and regional level in different countries. We have in previous research identified political, administrative and organizational motives behind this trend toward more project-based organizations within the public administration (Jensen, Johansson and Löfström, 2013). The problem is that the form of project organization carries inherent problems /special challenges when these projects are supposed to be implemented in permanent agencies and organizations. The purpose of this paper is to identify problems and challenges that public administrations face when projects are used as a form of organization in policy implementation, and identify possible strategies that can facilitate successful implementation when projects is used as a form of organization. The article takes its starting point in the policy implementation research and especially the seminal work of Richard Matland (1995) who bases the implementation analysis on the variables policy conflict and policy ambiguity as important factors by which it is possible to identify both various paradigms in implementation research, but also the factors that can explain the implementation results. This research tradition is complemented by research on temporary organizations. Our analysis shows that the use of project organization puts special demands on the players involved and if these are not taken into account, there is a high risk that projects which are designed to bring about social change risk becoming islands in the stream that are not producing the intended effects which policy makers and citizens expect.