This paper presents the results of a prospective study testing an expanded theory of reasoned action (TRA) to predict mammography participation. A questionnaire was developed to measure each of the expanded TRA model components. A sample was identified of 946 women age 40 and above who were invited to obtain a mammogram at the Group Health Cooperative of Puget Sound Breast Cancer Screening Program (BCSP). The sample was stratified by risk category as determined by the screening program. The study questionnaire was administered to all women in the sample within 2 weeks after they were sent the invitation to obtain a mammogram. Mammography participation was obtained from the BCSP data base 6 months after the invitation. Regression analyses found attitude, affect, subjective norm, and facilitating conditions to all be significantly associated with participation. The expanded TRA model explained 39% of the variance in women's intentions and 20% of the variance in participation behavior. A stepwise hierarchical regression found that no other psychosocial measures were able to improve the model predictions of behavior. An interaction between habit and intention was found such that women with larger numbers of previous mammograms were less likely to carry out their intentions than women with fewer previous mammograms. Contrary to expectations, some demographic characteristics did significantly improve prediction. The need for further work investigating the roles of fear and experience is discussed.