Little is known about levels of physical activity and attendance at phase IV community-based Cardiac Rehabilitation (CR) programs following completion of exercise-focussed, hospital-based phase III CR. This study aims to test, compare and combine the predictive utility of the Common-Sense Self-Regulation Model (CS-SRM) and the extended Theory of Planned Behaviour (TPB) with action planning for two rehabilitation behaviours: physical activity and phase IV CR attendance. Individuals diagnosed with coronary heart disease (n = 103) completed baseline measures of illness perceptions, intentions, perceived behavioural control (PBC), action planning and past physical activity in the last week of a phase III CR program, and 95 participants completed follow-up measures of physical activity and attended phase IV CR (objectively confirmed) 2 months later. Only one predictor (PBC/cyclical timeline) significantly predicted levels and change of physical activity. While illness perceptions were not predictive of phase IV CR attendance, the extended TPB model showed good predictive power with action planning and intention as the most powerful predictors. Amongst participants who planned when and where to attend phase IV CR at the end of phase III rehabilitation, 65.9% subsequently attended a phase IV CR program compared to only 18.5% of those who had not made a plan. This study adds to our understanding of cardiac rehabilitation behaviour after completion of health service delivered programs. Comparing theoretical models and rehabilitation behaviours contributes to the development of behaviour theory.