Individuals within social groups often show consistent differences in behaviour across time and context. Such interindividual differences and the evolutionary challenge they present have recently generated considerable interest. Social insects provide some of the most familiar and spectacular examples of social groups with large interindividual differences. Investigating these within-group differences has a long research tradition, and behavioural variability among the workers of a colony is increasingly regarded as fundamental for a key feature of social insects: division of labour. The goal of this review is to illustrate what we know about both the proximate mechanisms underlying behavioural variability among the workers of a colony and its ultimate consequences; and to highlight the many open questions in this research field. We begin by reviewing the literature on mechanisms that potentially introduce, maintain, and adjust the behavioural differentiation among workers. We highlight the fact that so far, most studies have focused on behavioural variability based on genetic variability, provided by e.g. multiple mating of the queen, while other mechanisms that may be responsible for the behavioural differentiation among workers have been largely neglected. These include maturational, nutritional and environmental influences. We further discuss how feedback provided by the social environment and learning and experience of adult workers provides potent and little-explored sources of differentiation. In a second part, we address what is known about the potential benefits and costs of increased behavioural variability within the workers of a colony. We argue that all studies documenting a benefit of variability so far have done so by manipulating genetic variability, and that a direct test of the effect of behavioural variability on colony productivity has yet to be provided. We emphasize that the costs associated with interindividual variability have been largely overlooked, and that a better knowledge of the cost/benefit balance of behavioural variability is crucial for our understanding of the evolution of the mechanisms underlying the social organization of insect societies. We conclude by highlighting what we believe to be promising but little-explored avenues for future research on how within-colony variability has evolved and is maintained. We emphasize the need for comparative studies and point out that, so far, most studies on interindividual variability have focused on variability in individual response thresholds, while the significance of variability in other parameters of individual response, such as probability and intensity of the response, has been largely overlooked. We propose that these parameters have important consequences for the colony response. Much more research is needed to understand if and how interindividual variability is modulated in order to benefit division of labour, homeostasis and ultimately colony fitness in social insects.