Objective To examine relative associations of dietary composition variables with body mass index (BMI; calculated as kg/m 2) among young and middle-aged US adults. We hypothesized that in subjects with physiologically plausible reported energy intakes, fiber intake would be inversely associated with BMI, independent of other dietary composition and sociodemographic variables. Subjects and Design Data from adults age 20 to 59 years in the Continuing Survey of Food Intakes by Individuals (CSFII) 1994–1996 were used. Exclusions were pregnancy or lactation, food insecurity, missing weight or height data, or having only one dietary recall. Based on our previously published methods, a physiologically plausible reported energy intake was calculated as being within ±22% of predicted energy requirements for the mean of two 24-hour recalls. Results Reporting plausibility ([reported energy intake÷predicted energy requirements]×100) averaged 83% in the total sample (N=4,539) and increased to 96% in the plausible sample (n=1,932). Only approximately 5% of the plausible sample consumed the Adequate Intake for fiber. In plausibly reporting women, fiber, its interaction with percentage energy from fat, and energy density were independently associated with BMI. Only percentage energy from fat was associated with BMI in men reporting plausibly. Compared with the total sample, stronger relationships between diet and BMI were observed among the plausible reporters. In women, a low-fiber (<1.5 g/MJ ⁎ ), high-fat (≥35% energy) diet was associated with the greatest increase in risk of overweight or obesity compared with a high-fiber, low-fat diet. Conclusions Weight control advice for US women should place greater emphasis on consumption of fiber.