Context: The leading cause of death in the U.S. is heart disease with obesity as the major correlate and, some believe, Cause: Obesity is caused by eating behaviors. The U.S. Preventive Services Task Force notes the need for, yet absence of, randomized controlled trials of behavior modification interventions for cardiovascular disease endpoints. Objectives: To modify habitual dietary behaviors with counseling, to assess retention of the resulting habits, and to compare cardiovascular effects of such modifications under diets differing in fat content. Design: One-year randomized controlled trial of three diet behavior interventions for cardiovascular and behavioral end-points with intermediate surrogate end-points enabling intent-to-treat analysis. Setting: Prairie States private cardiology practice. Participants: One hundred and twenty referrals screened for: cardiovascular co-morbidity, heart medication, and informed agreement. Intervention: One hour of diet protocol initial individual instruction and one half-hour of individual self-efficacy counseling every six weeks thereafter for one year. Main outcome measures: The primary cardiovascular measure was SPECT device measured physiologic coronary blood flow. Retention of modified habits was assessed with four-month post-study measurement of changes in Respiratory Quotient as a menu selection index and in Body Mass Index as a portion control index. Surrogate endpoints were weight, BMI, lipids and other standard blood chemistry indices. Results: Participant attrition, zero. Changes in coronary blood flow were correlated with diet protocol, R (95%CI)=0.48 to 0.69, p=2.4 x 10-12. Ischemia improved significantly under a lowered fat protocol and worsened significantly under lowered carbohydrate conditions. Surrogate endpoints, weight and blood chemistry, showed similar correlations, p<10-11. Four months post-study, lowered fat dieters had retained their intervention modified behaviors while lowered carbohydrate and vegan dieters had changed, R (95%CI)=0.53 to 0.74, p=2.5 x 10-14 for portion control and R (95%CI)=0.76 to 0.88, p<2.2 x 10-16 for menu selection.