Objective: To investigate the relationship between stunting and levels of overweight/obesity among South African school children, using two definitions of overweight and obesity, based on the WHO and International Obesity Task Force (IOTF) criteria.

Design: Cross-sectional descriptive analysis of the nutritional status of primary school children, using primary data from a rural community-based study undertaken in 1995 and secondary data from the South African National Primary Schools (SANPS) survey conducted in 1994. Stunting was measured according to the WHO definition of -2 Z scores height-for-age. Two sets of criteria were used to measure overweight and obesity—the WHO/NCHS standard based on the 85th and 95th centiles and the IOTF criteria.

Setting: The primary data source was from a rural KwaZulu-Natal community based survey. The secondary data source SANPS consisted of data at National and Provincial level; for this study only data from the province of KwaZulu-Natal was considered.

Subjects: Primary school children aged between 8 and 11 y of age; 802 from the primary data source and 24 391 from the secondary source.

Results: Moderate stunting ranged from 2.9 to 40.2%, and mild stunting ranged from 31.4 to 75%. The prevalence of overweight ranged from 0.4 to 13.3% (WHO criteria) and from 0.4 to 11.9% using the IOTF criteria; while obesity ranged from 0.1 to 3.7% (WHO) and from 0.1 to 1.5% (IOTF criteria). The prevalence of overweight and obesity was observed to be higher using the WHO definition than that of IOTF (0.050.55, in all cases). The levels of agreement in all cases were less for obesity than overweight (both ranged from 0.55 to 1.0). Females were observed to have higher kappa levels than their male counterparts; they also had higher prevalence levels of overweight and obesity across age and geographical group. Uniformly high levels of both mild and moderate stunting were observed both nationally and provincially. However, no excess relative risk of being overweight if stunted was observed in this study (P>0.05).

Conclusions: Caution must be applied when using either definition for obesity. However, very high levels of agreement occur for overweight. There is no obvious relationship between overweight and stunting in this study, but high levels of mild stunting were observed. Regular clinical and epidemiological monitoring of nutritional status needs to be undertaken in South Africa to examine possible future trends of overweight/obesity and their relationship with stunting, and for comparisons with global trends.