This thesis identifies and empirically validates the efficacy of psychological factors in the prediction of fitness club membership retention. Further, it seeks to address gaps in the literature created by three biases of the fitness club membership retention literature so far; a bias towards predicting renewal as opposed to cancellation, a bias towards predicting intentions as opposed to actual behaviour, and a bias towards conducting positivist research as opposed to pragmatist research. In response to these biases, this thesis focuses on cancellation (both intentional and actual) and adopts a pragmatic mixed methods approach. Firstly, a qualitative study was conducted with a sample of twenty-three current and ex-members of a fitness club which suggested a predictive role of four a priori themes (perceived service quality, perceived value for money, usage and brand identification) and five additional themes which were elicited during a template analysis (social identification, rapport, social physique anxiety, state anxiety and self-determination). These nine themes represented nine potential predictors of membership cancellation. A questionnaire was then developed, which measured these nine predictors and intentional cancellation, and was distributed to a large cross-section of current members (n=716) and a smaller cross-section of new members (n=89) which assessed the efficacy of the nine predictors in predicting intentional cancellation. In addition, actual cancellation data was collected twelve months after the questionnaire data was gathered for both studies. In relation to current members, state anxiety (in relation to staff) and intention to cancel, together significantly predicted actual cancellation. With regard to intention to cancel, whilst many of the predictors were predictive when analysed individually, when modelled together, only social physique anxiety, state anxiety (in relation to staff and members) and four self-determination sub-scales (external regulation, identified regulation, integrated regulation and intrinsic regulation) were predictive of intentional cancellation. In relation to new members, actual cancellation could not be predicted. With regard to intention to cancel, overall perceived service quality and three brand identification scales; brand attractiveness, brand prestige and brand distinctiveness were found to be predictive of intentional cancellation. The findings of the studies are integrated and discussed, and suggestions are made regarding future research directions and implications for practice.