Cyclical stepping (gait) has been studied extensively. Some of these results are reflected in the straight and curved path step-following algorithms in commercial digital human modeling (DHM) implementations. With the aid of these algorithms, DHM users define start, intermediate, and end path points and the software generates a walking-like motion along the path. Most of these algorithms have substantial limitations, among them that the figures exhibit “foot skate,” meaning that the kinematic constraint of foot contact with the ground is not respected. Turning is accomplished by pivoting the entire figure, rather than through realistic lower-extremity motions. The simulation of the non-cyclical stepping motions accompanying manual material handling pickup and delivery tasks requires manual manikin manipulation. This paper proposes a paradigm for the simulation of stepping behavior in digital human models based on a model of foot placements and motions. Cyclical and non-cyclical transition stepping behaviors are handled with the same structure, allowing for smooth transitions between gait and non-cyclical behaviors. The model is based on a laboratory experiment in which participants used one or both hands to move heavy and light loads between shelves located in various positions one to five steps apart. The data were used to develop a model of the transition stepping that occurs at load pickup and delivery.