Most longitudinal, correlational studies on health-behaviour change examine effects of Time1 social-cognitive predictors on subsequent behaviour. In contrast, our research focuses on associations between changes in predictors with change in behaviour. The Health Action Process Approach (HAPA) distinguishes between motivational predictors for intention formation and volitional predictors for behavioural change and served as theoretical basis. Two online-studies were launched targeting different behaviours (low-fat diet, smoking), different samples (Study 1: N = 469; Study 2: N = 441) and different time spans (Study 1: 3 months, Study 2: 4 weeks). Data were analysed by means of structural equation modelling with latent difference scores. Both studies resulted in almost parallel prediction patterns. Change in risk awareness and change in outcome expectancies did not result in change in intentions, whereas change in self-efficacy was of crucial importance. Change in behaviour was associated with change in action planning and action control over and above the effects of intentions. In one study, increases in self-efficacy yielded increases in behaviour change. Results demonstrate that change in action planning and especially action control was of great importance for behaviour change across two different behaviours. Analysing change in social-cognitive predictors allows drawing precise conclusions for interventions.