Irritable bowel syndrome (IBS) is a common disorder of gut-brain interaction characterised by symptoms of abdominal pain, occurring at least 1 day per week, and a change in stool frequency or form. Individuals with IBS are usually subtyped according to their predominant bowel habit, which is used to direct symptom-based treatment. However, this approach is probably an oversimplification of a complex and multidimensional condition, and other factors, such as psychological health, are known to influence symptom severity and prognosis. We have previously used latent class analysis, a method of mathematical modelling, to show that people with IBS can be classified into seven unique clusters based on a combination of gastrointestinal symptoms, abdominal pain, extraintestinal symptoms, and psychological comorbidity. The clusters can be used to predict the prognosis of IBS (eg, symptom severity), health-care use (eg, consultation behaviour, prescribing, and costs), and impact (eg, quality of life, work and productivity, activities of daily living, and income). These clusters could also be used to increase the personalisation of IBS treatment that better recognises the heterogenous nature of the condition. We present new data providing additional validation of our seven-cluster model and conduct a comprehensive evidence-based review of IBS management. Based on this evidence, we propose a framework of first-line and second-line treatments according to IBS cluster. Finally, we discuss what further research is needed to implement this approach in clinical practice, including the need for randomised trials comparing cluster-based treatment with conventional treatment according to stool subtype.