BACKGROUND. The Women's Health Trial (WHT) was a feasibility study for a randomized controlled trial designed to test the hypothesis that a reduction in dietary fat reduces breast cancer incidence among women age 45 to 69. Between 1984 and 1988, 2,064 women participated in its two phases. METHODS. A follow-up study of 525 women who were randomized to receive the WHT dietary intervention program was conducted to assess maintenance of the diet 1 year on average after the trial ended. Among 448 participants, the mean percentage of energy from fat as measured by a food frequency questionnaire was 40.0% at baseline, 26.3% at the end of the trial, and 27.7% at follow-up. Based on 408 women with complete data, a recursive model was estimated, describing the influence of baseline characteristics of the women on attendance at intervention program sessions, adherence to the diet during the trial, and long-term maintenance of the diet after the trial ended. The effects of women's experiences during the trial on adherence and long-term maintenance were investigated as well. RESULTS. Attendance at the educational sessions was strongly related to adherence to the diet during the trial (P less than 0.001), and adherence was the most important predictor of long-term maintenance (P less than 0.001). The percentage of energy from dietary fat at baseline was an important correlate of both adherence (P less than 0.001) and long-term maintenance (P less than 0.001). College-educated women were more likely to adhere to the diet during the trial (P less than 0.001). Feelings of deprivation adversely affected long-term maintenance (P less than 0.01), primarily through their effect on adherence during the trial (P = 0.01). Costliness of the diet in time and money negatively influenced long-term maintenance (P less than 0.05). Development of a distaste for fat encouraged adherence (P = 0.06). CONCLUSIONS. The low-fat dietary pattern established during the WHT was maintained for as long as 20 months after the trial ended. A recursive model was useful in analyzing the process and correlates of long-term maintenance of dietary behavior change. Both predisposing variables and women's experiences while on a low-fat diet were associated with long-term maintenance. The results suggest that feelings of deprivation should be avoided, perhaps by use of low-fat substitutes, by those attempting to lower their dietary fat and that more research is needed on the development of a distaste for fat among individuals who adopt low-fat diets.