Stunting is a useful marker of poor child growth and development, and is associated with long-term consequences for cognitive development, health, and adult earnings. However, measuring growth faltering is conceptually and methodologically challenging. Nationally representative surveys have produced a wide range of estimates of stunting prevalence among children under 5, complicating the tracking of trends over time. The paper examines the difficulties associated with measuring stunting and factors that may contribute to variation in stunting estimates. A comparison of recent South African surveys reveals methodological and sampling differences, as well as indications of possible data quality issues. These issues limit the comparability of stunting estimates across surveys. It concludes with recommendations to improve stunting measurement to allow better monitoring of progress.