There is only one set of estimates of the prevalence of chronic undernutrition world-wide. These estimates emanate from the FAO and are claimed to be comparable across countries. In this paper it is demonstrated that the FAO has used an estimation model (and data) that induce bias in the estimated overall prevalence of undernutrition. The bias is found to be large and to compromise the comparability of undernutrition across countries. Since the international community’s efforts to reach the recently agreed objective to reduce by half the number of undernourished before the year 2015 are to be directed - and monitored - with the help of the FAO estimates, these must be thoroughly improved. Suggestions for improvements are offered in the paper.