A dietary change intervention based on Gollwitzer's (1993) concept of implementation intentions was used to see whether transitions between transtheoretical model stages could be induced in a field experiment. Participants (N=554) were randomized to receive either questionnaire-only or questionnaire-plus-implementation intention. Results showed that, controlling for variables from the theory of planned behavior, implementation intentions helped a significantly greater number of people to progress through the stages of change than did receiving the questionnaire alone. However, the implementation intention intervention did not prevent regression from the stages of change. The discussion focuses on the importance of targeting implementation intention interventions and on the future use of health professional-led analyses of critical situations and precise identification of appropriate responses to prevent regression.