This briefing paper critically examines two concepts that permeate contemporary policy debates about the advisability and feasibility of introducing comprehensive social protection programmes in low-income countries – ‘dependency’ and ‘graduation’. Both issues are commonly raised by governments and donors that are sceptical about making firm, long-term commitments to social transfer programmes. ‘Dependency’ is generally thought of as a negative but inevitable consequence of providing people with regular social transfers on a long-term basis. The concern is that beneficiaries will come to regard these transfers as an alternative means of meeting basic consumption needs, and will lose any motivation to secure their livelihoods through their own efforts. ‘Graduation’ is often presented as a positive antidote to dependency; financial assistance to poor individuals and families in distress should be limited in scale and time in order to avoid the ‘dependency trap’, and complementary programmes should be put in place to ensure that beneficiaries are able to ‘graduate’ from ‘handouts’ and become self-reliant. However, this paper shows that it is not difficult to expose flaws in these generalisations about human behaviour and the risks of social transfers.