Governments and organizations invest huge sums of money in development interventions to explicitly address poverty and its root causes. However, a high proportion of these do not work. This is because interventions are grounded in flawed assumptions about how change happens -- change is rarely linear, yet development interventions are almost entirely based on linear planning models. Change is also characterized by unintended consequences, which are not predictable by planners and by power dynamics, which drive outcomes towards vested interests rather than real need. Development processes need to engage effectively with these sorts of complex system dynamics. This book provides a conceptual framework for this thinking, offers detail case studies of interventions which have been built on this philosophy and which demonstrate key facets of it. It articulates some clear methodological underpinnings for this work, and draws out the implications both for development, practice and practitioners.