The study was made to assess the order of error and the sources of inaccuracy of anthropometric measurements made by a team of 9 trained observers. Some observers displayed internal inconsistencies in making the measurements, and the variability in overall means due to differences between observers were relatively large. Restricting certain observers to the instruments upon which they show consistency and additional training of these observers, or the use of control charts based on tolerance limits, estimated from “pooled” variances of intra- and inter-individual variances, or intra-individual variances alone, would reduce the present errors to more reasonable levels.