Objective: To de?ne a de facto reference body mass index (BMI) for women in developing countries and compare its performance with the Quetelet BMI. Design: A logarithmic equation for elite wt/ht references was developed using the weight (wt) and height (ht) of 10 524 non-pregnant elite mothers. Functional outcomes were compared using both BMIs. Setting: Forty-six national surveys from 36 developing countries. Subjects: Mothers 15 –49 years old. Results: The de facto reference standard deviation showed 2.2% of elite women were undernourished and 6.3% overnourished; lower and upper limits for ‘ideal’ wt/ht were 73% and 137%, respectively. Compared with the age-based and medium frame standards of the First and Second National Health and Nutrition Examination Surveys (NHANES I & II), the de facto reference de?ned fewer women as undernourished (5.3% vs. 10.5% and 14.4%, respectively), but more as overnourished (20.4% vs. 13.7% and 15.9%, respectively). In the de facto reference, BMI ? wt=ht 1:6 . Using the agebased and medium-frame-based Quetelet BMI (wt/ht 2 ), 28.5% and 31.7% had a low and 13.0% and 14.7% a high BMI, respectively. For the de facto reference, 18.4% of the elite mothers had a low BMI and 19.3% a high BMI. Applying the de facto reference to all women showed that the distribution of BMI was similar irrespective of the reference used. Functional outcomes were similar for both BMIs. Conclusions: The NHANES I & II growth curves de?ne more women as overnourished than the de facto curve, but the opposite for de?ning the undernourished. Functional outcomes were similar for both BMIs, suggesting there is no great advantage to using a de facto BMI based on national-level data from these 46 national surveys.