Donald Broadbent's work in developing and validating his Cognitive Failures Questionnaire (CFQ) incidentally contributed to our understanding of what self‐assessment questionnaires (SAQs) can tell us about the nature of cognitive changes that occur in old age, and their effects on people's everyday lives. The literature on use of SAQs in age comparisons is reviewed in the context of this work, as is new empirical evidence that older people complain of general loss of memory efficiency but, paradoxically, report fewer lapses than young adults on the CFQ and other SAQs. These new data provide a basis for discussion of the reliability and consistency of SAQs and of their validity as tools to explore a wide range of changes associated with cognitive ageing: objective changes in cognitive ability and in objective functional competence; changes in people's understanding of their own cognitive processes, in their knowledge of cognitive and mnemonic strategies, and in their abilities to make use of these strategies; the extent to which responses given by older and younger adults reflect emotional and personality variables such as depression, anxiety, neuroticism, extraversion, and perceived locus of control. Self‐reports by elderly adults on the CFQ, and other SAQs, simultaneously provide so many different kinds of information about the changes they experience, and about their attitudes and adaptations to old age, that apparently straightforward analyses give misleading answers. Methodological precautions are suggested for determining which questions can, and are, being asked and which analyses may, and which cannot, answer them.