Human figure models included with today’s high-end simulation systems are widely used for ergonomic analysis. Male and female figures can be scaled by inputting standard anthropometric dimensions, such as stature, body weight, and limb lengths to represent a large range of body sizes. However, the shapes of these scaled figures are often unrealistic, particularly for extreme body types, because only a relatively few dimensions of the figures are adjusted. Several software packages have the capability of adjusting the figure to match a body surface scan of an individual, but this capability is insufficient when individuals with a wide range of attributes must be simulated. We present a new figure generation capability in the Jack human modeling software based on a statistical analysis of laser scans of hundreds of men and women. A template mesh based on the Jack figure model was fitted to each scan, and a statistical shape model was created using principal component analysis and regression. The model was integrated within the Jack software to create Jack figures with highly realistic body shapes, usable natively with all simulation and analysis features.