The aim of this article is to identify some of the difficult areas in researching the South African social security system, and to offer suggestions as to the kind of further research that is needed to improve our understanding over time. The article starts with a brief ‘tour’ through some of the articles dealing with the subject that have already appeared in Development South Africa, with a view to showing their relationship to each other and to the main data sets which have been used. This is followed by sections on the size of pension income relative to other sources of income; the use of the term ‘transfer income'; the relationship between policy design and socio-economic behaviour; and some difficulties when comparing different government expenditures on welfare. It is hoped that this will contribute to a more careful analysis and interpretation of existing research, as well as the more precise formulation of social security research in future.\n\n