The aim of this study was to compare how different diagnostic criteria can influence the estimation of obesity in children. Five hundred and eighty-seven children from Southern Italy were evaluated for the presence of obesity according to six different methods: two using Ideal Body Weight, according to Tanner and the National Center for Health Statistics, and four using Body Mass Index, according to Rolland-Cachera, Must, Cole (International Obesity Task Force) and the Centers for Disease Control and Prevention. Large discrepancies were found between old and new methods in identifying childhood obesity with respect to absolute prevalence, differences between the sexes, and age-related trends. The use of different weight-height indices, the employment of different cut-off points, the large differences between the reference populations, and the different time periods of data collection explain differences between methods. This should be always taken into consideration when data from different epidemiological or clinical studies are compared.