We examined the repeatability of near-infrared interactance (NIR) body fat determination as compared with that of body mass index (BMI), waist-to-hip ratio (WHR), and waist girth. Thirty-nine volunteers (16 men, 23 women) had percent body fat (%BF) measurements made with a portable NIR device as well as the standard anthropometric indices of height, weight, waist girth, and hip circumference. Frame size and physical activity levels were also determined. For each participant, three independent measurements of each index were made by two trained readers during a 2-week period. The two readers varied significantly in their measurement of %BF and hip circumference. The variability in %BF was largely due to differences between the first and the second measurements, and only for one of the readers. Second and third measurements were not statistically significantly different for either reader, suggestive of a training effect. Variance component calculations revealed that the reliability of NIR is 95.3%, compared with 99.9% for BMI; 93.4% for waist girth; and 82.4% for WHR, with the majority of the remaining variance accounted for by the method itself. We conclude that the NIR method has good repeatability, with low intra- and interobserver variability, provided that readers are carefully trained. However, the NIR device offers little advantage in reliability over conventional measures of adiposity such as waist girth or BMI, and requires additional input of weight, height, frame size, physical activity level, age, and gender data to calculate %BF. Associations of NIR and other anthropometric indices with cardiovascular risk factors in this population will provide additional insight into the merit of NIR body fat assessment.