There is some evidence for a positive relationship between affective responses to exercise and future physical activity behaviour, however previous research has been laboratory-based. This thesis aimed to contribute to the current theoretical understanding of affective responses to outdoor physical activity (walking and running), and the relationship of these affective responses to physical activity behaviour change, via three separate empirical studies and a synthesis of the qualitative data across studies. This research makes a number of contributions towards our understanding of the relationships between affective responses to exercise and future physical activity behaviour. Firstly, there was no reliable evidence to indicate that affective responses during everyday outdoor walking at baseline predicted future physical activity behavior (Study 1), though social factors were found to be important in both quantitative and qualitative analyses. Qualitative results suggested that the meanings of everyday walking might be one potential reason for this. Using a run-walking session as an exercise stimulus, affective responses mid-session, end-session and post-session did predict changes in physical activity six months later, whereas early-session affective responses were not significantly related to future physical activity (Study 2). Using results from the first two studies, a beginner run-walking intervention based on affect-regulated exercise intensity in was developed and optimised (Study 3), which was shown to be acceptable, enjoyable and feasible for a future randomised trial testing adherence. Qualitative results from the empirical studies were synthesised, and commonalities, differences and new insights into affective responses to walking and running were found. Results included themes of sources of affective responses (environment, social factors, dis/comfort, comparison with previous self and stages of exercise) and psychological outcomes of exercise (achievement, emotional and mental outcomes). The affective measure used in all empirical studies was analysed, with results showing the most promise for the measure in social situations.