Objectives. The predictive validity of the theory of planned behaviour is well established, but much less is known about: (a) whether there are causal relationships between key components of the model and (b) how to go about changing the theory of planned behaviour variables. This study tested the ability of outcome and process simulations to change variables specified in the theory of planned behaviour in relation to blood donation. Design. Participants (N = 146) were randomized to one of four conditions: outcome simulation only, process simulation only, process-plus-outcome simulation and a distractor control condition. The dependent variables were state anxiety, and intention attitude, subjective norm and perceived control from the theory of planned behaviour. Methods. Participants were asked to empty their mind and visualize themselves: (a) after donating blood (outcome manipulation), (b) preparing to donate blood (process manipulation), (c) both preparing to donate blood and after having donated blood (process-plus-outcome manipulation) or (d) both preparing to get a high mark and after having got a high mark on their course (control condition). Following mental rehearsal, participants completed the dependent variables. Results. There were no main effects of outcome simulation, but process simulation successfully increased intention, subjective norm and perceived control. There was also a significant outcome simulation x process simulation interaction for attitude. The effect of the process manipulation on intention was mediated by subjective norm and perceived control. Conclusions. The findings show promise for the use of mental simulations in changing cognitions and further research is required to extend the present findings to other health behaviours. © 2008 The British Psychological Society.