There is now good evidence to show that family therapy is effective in helping families with a wide range of presenting problems (Carr, 2014a; Shadish & Baldwin, 2003; Stratton, 2011). Although family therapy has a strong evidence base, much of the focus in the research literature has been on outcomes; far fewer studies have attempted to investigate the process of change in family therapy, and there has been a call for greater research in this area (Heatherington et al., 2005; Sexton & Datchi, 2014; Vilaça, Margarida, & Ana Paula, 2014). The inability to evidence how the process of family therapy works leaves it open to criticisms concerning credibility. Since family therapy is one of the few alternatives to one-to-one talking therapies (such as cognitive behavioural therapy) (Stratton & Lask, 2013), its devaluing would limit client choice. In order to justify its position as an alternative to one-to-one ways of working, family therapy needs to be able to evidence the link between theories of change, the process of change and outcomes. The present study attempts to address some of the above concerns by exploring how family therapists in the U.K understand the change process in family therapy both in theory, and in their practice. In addition, it explores how the rationale that therapists’ provide for their interventions in specific therapy sessions relates to how families conceptualise and experience change in the same sessions. A multi-modal triangulation methodology was used, whereby eight family therapists were interviewed in prospective qualitative interviews, and then two families and two family therapy teams were interviewed while they reviewed therapy sessions using the process methodology of Inter-personal process recall (IPR). The two methodologies revealed several themes. Three super-ordinate themes emerged from the prospective interviews: ‘safe-space’, ‘perspective taking’ and ‘privileging the change’. The IPR interviews yielded four super-ordinate themes: two from family therapy teams; ‘expressing a clear rationale’ and ‘linking theory and process is difficult’ and two from families ‘things we found helpful’ and ‘things that we didn’t like’. Convergence and divergence between the data sets are discussed as well as implication for further research and limitations of the current study.