Little is understood about the patterns in which frequent attenders consult primary care over time, or about the clinical or behavioural reasons for doing so. The aim of this study was to use a sorting task to identify the important features of patterns of attendance to general practitioners generated by frequent attenders over a period of 44 months. A total of 12 experts performed a sorting task on consulting patterns of the 100 most frequent attenders from one general practice in Leeds. A similarity matrix was constructed which was then subjected to cluster analysis. A range and diversity of patterns was clearly demonstrated within the data. The determining factors of the groupings by the experts included frequency, regularity, bursts of consultations, patterns of consulting after bursts, periods of not consulting, and seasonal or annual trends. The cluster analysis provided six final clusters, the largest of which had no fixed or classifiable pattern of attendance. The other five clusters were ‘regular without bursts’, ‘supernovas’, ‘gap and prolonged burst’, ‘bursts and sporadic attending’ and ‘regular as clockwork’. Distinct patterns of frequent attendance exist, and it seems likely that the clinical and behavioural reasons underpinning such behaviours differ. This work furthers our understanding of the way in which frequent attenders consult, and the results have implications both for general practice and for future research. The methodology described may have applications in other areas of primary care and biomedical research.