Aim and Research Questions Adopting a consumer’s behavior marketing approach, the study questions loyalty to a sporting club in the case of horse riding. In a national context of a recent drop in the number of FFE (French Equestrian Federation) members, there is today a need to better understand riders' relationship with their equestrian centre. In the literature, although many researches have focused on loyalty, the context of recreation-related services remains under-investigated. In this context, this study aims to understand mechanisms of riders’ loyalty to equestrian centres. Theoretical Background In line with much research on loyalty (Lichtle, Plichon, 2008), we first hypothesize that satisfaction increases loyalty (H1). Algesheimer et al. (2005) showed that community commitment has a positive effect on loyalty, and that community commitment increases if members are participating in collaborative activities and are willing to correspond to community actions. In the case of horse riding, we note that all riders express a need for "belonging to a community" sharing the same passion (Keaveney, 2008). Consequently, we suppose that community commitment increases satisfaction (H2) and that commitment to friends’ group increases community commitment (H3a) and satisfaction (H3b). Riders look for a relationship with their teacher (Klisch, 2009). We thus hypothesize that affective commitment to the teacher has a positive effect on satisfaction (H4a), on community commitment (H4b), and on loyalty (H4c). We also assume that affective commitment to the teacher increases switching costs (H4d), in line with the positive switching costs highlighted by Lichtle and Plichon (2008). Switching costs are assumed to influence negatively satisfaction (H5a) and loyalty (H5b), as well as alternative offer attractiveness (H6a and H6b) 2 (Lichtle, Plichon, 2008). Finally, we assume age-differentiation in the model (H7), based on Casper and Stellino's work (2008) who confirmed the need for age-related differentiation in sport commitment analysis. Research Design, Methodology and Data Analysis First, we conducted a qualitative exploratory study based on semi-structured interviews with 30 riders in various equestrian structures and various areas. Based on these results, we realized an online quantitative survey in France among 630 respondents. These ones are from 15 to 77 years of age, with an over- representation of women (92%) (as in FFE members (83%)), and of higher socio- economic status (39%). Average time of horse riding practice is 12.8 years. Secondly, we measured the constructs of loyalty, satisfaction, switching barriers, affective commitment and community commitment with scales based on 2 to 4 items (through 5 point Likert scales), found in the literature. Exploratory and confirmatory factor analysis, as well structural equation modeling, have been processed with R software. Finally, demographic determinants were chosen according to life cycle, as in Casper and Stellino's work (2008), to test a multigroup model: teenage period 15 to 18 y.o; student or active life entry period 19 to 25 y.o.; young active and family period 26 to 36 y.o.; and 37 y.o. and over when you evolve more sedentary. Results and Discussion The overall model verifies convergent validity (AVE > 0.5 for all constructs), reliability (alpha > 0.7 for all constructs except one equal to 0.66), discriminant validity (AVE> square of the correlations for all constructs), and goodness of fit indices (CFI=0.904; RMSEA=0.069). All hypothesis are validated except H3b. The results highlight that affective commitment to the teacher influences client satisfaction, community commitment, perceived switching costs and loyalty, supporting the major influence of horse-riding teachers on customers’ loyalty to equestrian centres. Turning now to the multigroup analysis, metric invariance is verified (p=0.983) whereas structural invariance is not (p=0.000), confirming the moderating role of age. All statistical properties of the models are verified (details available upon request). Interestingly, one major difference between groups concerns the affective commitment to the teacher, which increases loyalty through satisfaction for all groups, except for experienced riders. For these ones, satisfaction has no effect on loyalty, and commitment to the teacher has a direct effect on loyalty. 3 Theoretically, our results show that the relationship between satisfaction, commitment(s) and loyalty may differ depending on demographic variables. However, further analysis should include other antecedents of commitment, such as perceived value of the centre or position involvement (Bodet, 2012). Marketing implications suggest the key role of the teacher in the relationship with the structure for all market segments. Turning to the young active segment, which corresponds to a critical period for loyalty towards equestrian centres, equestrian centres should use teacher’s influence on community commitment to increase young actives’ loyalty. Conclusion First action levers for equestrian centres begin to emerge, to better retain and satisfy their customers. These levers remain to be deepened and adapted according to riders’ profile. Finally, this analysis based on our model of loyalty in equestrian centre underlines the need to adapt management strategies in marketing and teaching according to recreational riders ’age.