Background Community-based multi-chronic illness exercise programmes provide a resource efficient way to assist with the treatment/management of chronic illness. Despite the well-documented benefits of participation in community-based chronic illness exercise programmes (CCEP), adherence remains sub-optimal with positive effects of such programmes relying on optimal adherence. Difficulty remains in interpreting rates of adherence and correlates due to programme heterogeneity. Given the health benefits to be gained from regular attendance, empirically based interventions and strategies to optimise adherence are warranted. The purpose of this thesis was to i) identify rates of initiation, attendance and retention and to identify factors associated with adherence to a CCEP and ii) design, implement and evaluate the feasibility of an intervention aimed at optimising adherence. Methods Individuals with chronic illness were referred to a CCEP by health professionals. Study 1 (n=381) was a prospective study encompassing cross-sectional and prospective analysis designs to identify the rates and correlates of adherence. Primary statistical analysis included independent samples t-tests, bivariate correlations, logistic regression and negative binomial regression. An intervention based on Study 1 and the existing evidence base was designed and developed using the behaviour change wheel. Study 2 (n=295) was a feasibility study of a behaviour change intervention aimed at optimising adherence rates (composed of intervention at baseline, week-4 and 12). Study 2 also utilised a quasi-experimental research design to compare trends in rates of adherence over 24 weeks between those receiving the intervention and those not. xvii Results Study 1: The initiation rate was 73% and the mean number of sessions attended in the 24-weeks was 11. Bivariate analysis identified intentions, self-efficacy and BMI as correlates of initiation. Following adjusted analysis, the associations were no longer statistically significant. Further bivariate analysis indicated chronic illness group, distance to facility, intentions, self-efficacy for exercise and attendance in the first 4 weeks as correlates of 24-week attendance. In an adjusted analysis, age and distance to facility were the only as predictors of attendance. The first 4 weeks’ attendance when added to the previous models was the only independent predictor of 24-week attendance. Study 2: A theoretically informed behaviour change intervention was designed utilising the behaviour change wheel. The intervention was evaluated in terms of its feasibility, and with only 25.6% of participants attending all component of the intervention as planned it is deemed not feasible in its current form. The intervention was also evaluated using intention to treat and per-protocol analysis and using an intention to treat analysis trends indicate that retention was greater in the Intervention than the Standard group. There was no difference in initiation or attendance between the Intervention and Standard group. Per-protocol analysis found trends toward improvements in initiation (96% vs 73.2%,), attendance (mean  SD) (22.22 ±12.59 vs 11.13±12.42) and retention (56.3% vs 28.7%) in the Intervention compared to the Standard group. Conclusion: Bivariate correlates of adherence to a CCEP are diverse, with early stage attendance the strongest predictor of 24-week attendance. The inclusion of a xviii theoretically informed behaviour change intervention can improve adherence in those who attend all components of the intervention, however, the intervention did not prove feasible. Further research should identify correlates of adherence to CCEP to inform the design and implementation of interventions to optimise physical and mental health in men and women with chronic illness.