The international (International Obesity Task Force; IOTF) body mass index (BMI) cut-offs are widely used to assess the prevalence of child overweight, obesity and thinness. Based on data from six countries fitted by the LMS method, they link BMI values at 18 years (16, 17, 18.5, 25 and 30 kg m(-2)) to child centiles, which are averaged across the countries. Unlike other BMI references, e.g. the World Health Organization (WHO) standard, these cut-offs cannot be expressed as centiles (e.g. 85th). To address this, we averaged the previously unpublished L, M and S curves for the six countries, and used them to derive new cut-offs defined in terms of the centiles at 18 years corresponding to each BMI value. These new cut-offs were compared with the originals, and with the WHO standard and reference, by measuring their prevalence rates based on US and Chinese data. The new cut-offs were virtually identical to the originals, giving prevalence rates differing by < 0.2% on average. The discrepancies were smaller for overweight and obesity than for thinness. The international and WHO prevalences were systematically different before/after age 5. Defining the international cut-offs in terms of the underlying LMS curves has several benefits. New cut-offs are easy to derive (e.g. BMI 35 for morbid obesity), and they can be expressed as BMI centiles (e.g. boys obesity = 98.9th centile), allowing them to be compared with other BMI references. For WHO, median BMI is relatively low in early life and high at older ages, probably due to its method of construction. © 2012 The Authors. Pediatric Obesity © 2012 International Association for the Study of Obesity.