This paper advances a method of extrapolating, to target populations, inter-relationships in the anthropometry of an existing population database. Previous methods, including those that are based on using proportionality constants and those that involve developing regression relations, make use of stature as the primary predictor. This new approach distinguishes itself by accounting for the variability, across all the measures, that is not correlated with stature or other anthropometry. It builds on previous efforts by incorporating both stature and body mass index (BMI) as basic predictors in a single-step regression analysis of existing anthropometric data. The method is validated and shown to produce anthropometric measures for a population that are equivalent to the true measures. Additionally, this paper examines the effectiveness of multi-step regression in predicting anthropometry. This technique is compared with the single-step regression approach and is shown to not always result in improved accuracy. While the methodology proposed in this paper is not a replacement for gathering true anthropometric data from populations of interest, it is a useful tool for estimating larger sets of anthropometry when only a few measures (such as stature and BMI) are available. This will facilitate the use of digital human models in designing, with increased efﬁcacy, for human variability.