Aims to explore, using qualitative methods, the concept of exercise dependence. Semistructured interviews were undertaken with subjects screened for exercise dependence and eating disorders. Female exercisers, four in each case, were allocated a priori to four groups: primary exercise dependent; secondary exercise dependent, where there was a coincidence of exercise dependence and an eating disorder; eating disordered; control, where there was no evidence of either exercise dependence or eating disorder. They were asked about their exercise and eating attitudes and behaviour, as well as about any history of psychological distress. Their narratives were taped, transcribed, and analysed from a social constructionist perspective using QSR NUD*IST. Participants classified as primary exercise dependent either showed no evidence of exercise dependent attitudes and behaviour or, if they exhibited features of exercise dependence, displayed symptoms of an eating disorder. Only the latter reported a history of psychological distress, similar to that exhibited by women classified as secondary exercise dependent or eating disordered. For secondary exercise dependent and eating disordered women, as well as for controls, the narratives largely confirmed the a priori classification. Where exercise dependence was manifest, it was always in the context of an eating disorder, and it was this co-morbidity, in addition to eating disorders per se, that was associated with psychological distress. As such, these qualitative data support the concept of secondary, but not primary, exercise dependence.