This paper reviews recent research on take-up with a view to identifying the most promising lines of further enquiry. In the late 1970s take-up research seemed to be foundering in the face of the complexity of the factors affecting claiming decisions. Progress came via the development of models of the claiming process. The most influential postulated a series of thresholds which claimants must pass in a set sequence on the way to claiming a benefit. One question for future research is whether this model is still the best starting point or whether there are others which capture the key factors and their interaction more fully and accurately. Another approach is to look at claiming decisions indirectly by applying multivariate techniques to continuous survey data to examine the relationship between variables such as age, income and housing tenure and differing probabilities of claiming. A second question is what the two approaches can offer each other. Overhanging all the issues facing take-up research is the problem of sampling a population—eligible non-recipients—for which, almost by definition, no suitable sampling frame exists. The paper concludes by examining various options for identifying this elusive group.