<p>The role of the therapeutic alliance (TA) has largely been ignored in the field of high-risk violent offender treatment. The focus on effective manualised treatment that reduces recidivism has led to improvements in treatment programme delivery, but at the cost of examining the therapy process. Considering previous research has consistently linked levels of alliance with treatment outcome in clinical and community treatment settings, it is imperative to investigate the role of the TA within high-risk violent offender treatment, because of the particularly challenging group that they represent. The aim of this thesis was to examine the relationship between the TA and treatment outcome, and the various factors that influence this relationship, within a violence prevention setting. To achieve this aim, a longitudinal study was conducted at the Rimutaka Violence Prevention Unit (RVPU) in Wellington, New Zealand, with a cohort of 70 men in treatment and their therapists, examining the TA, treatment outcome and associated variables over four time points throughout the eight month treatment programme. The results of this research are reported as four related studies. Study One explored the structure and patterns of the Working Alliance Inventory (WAI). Study One Part A was a Confirmatory Factor Analysis of the WAI, which tested the competing models of the factor structure of the WAI and explored whether rater perspective (client, therapist, observer) had an effect on the structure. It was found that a two-factor structure was the best fit for the WAI, and that all rater perspectives shared this structure. Study One Part B explored the pattern of the WAI over the four time periods of this study in order 8 to understand how the WAI changes over time, and whether this pattern differed by rater perspective. The results confirmed that changes in all rater perspectives showed a linearly increasing pattern of alliance over time. Study Two explored the client factors that affect the initial formation of the TA and examined whether these factors were specific to an "offender" or "general" client profile informed by previous research. Two client factors specific to an offender profile - motivation to change and criminal attitudes - were found to be significantly associated with the initial formation of the TA. Study Three examined the relationship between the TA and treatment outcome, and explored whether there were any factors that co-varied with or moderated this relationship. A small but significant association between alliance and outcome was found; however no significant co-varying or moderating factors were discovered. Lastly, Study Four drew together the data from Study Two and Study Three and tested whether these results fit the Revised Theory of the Therapeutic Alliance (RTTA) model (Ross, Polaschek, & Ward, 2008), or other models previously reported in the literature. Several significant models were found that partly supported the RTTA. The best of these models incorporated client motivation to change, TA and treatment outcome as measured by change in risk of violent reoffending. Overall, the results of this study support the importance of the TA and client motivation to change in violent offender treatment. The implications for these results and the clinical applications are discussed, limitations are outlined, and directions for future research are suggested.</p>