Anthropometric indices of adiposity include BMI, waist circumference and waist:height ratio. In the recruitment phase of a prospective cohort study carried out between 1998 and 2002 we studied a population sample of 11 786 white Caucasian non-pregnant women in Southampton, UK aged 20-34 years, and explored the extent to which proposed cut-off points for the three indices identified the same or different women and how these indices related to adiposity. Height, weight and waist circumference were measured and fat mass was estimated from skinfold thicknesses; fat mass index was calculated as fat mass/height1.65. Of the subjects, 4869 (42 %) women were overweight (BMI > or = 25 kg/m2) and 1849 (16 %) were obese (BMI > or = 30 kg/m2). A total of 890 (8 %) subjects were not overweight but had a waist circumference > or = 80 cm and 748 (6 %) subjects were overweight but had a waist circumference < 80 cm (6 %). Of the women, 50 % had a BMI > or = 25 kg/m2 or a waist circumference > or = 80 cm or a waist:height ratio > or = 0.5. Of the variation in fat mass index, 85 % was explained by BMI, 76 % by waist circumference and 75 % by waist:height ratio. Our findings demonstrate that many women are differentially classified depending on which index of adiposity is used. As each index captures different aspects of size in terms of adiposity, there is the need to determine how the three indices relate to function and how they can be of use in defining risk of ill health in women.