Foot placements are critical determinants of musculoskeletal loading in manual materials handling (MMH) tasks because of their effects on torso and upper-extremity posture. In spite of this importance, current ergonomic assessment tools lack valid models to predict the foot locations workers will chose. To address this need, a new comprehensive approach for modeling stepping behavior in MMH tasks was developed, based on field observations and a laboratory study. A method for qualitatively describing stepping strategies, the Lexical Transition Classification System (L-TRACS), was developed from observations of experienced operators performing MMH transfer tasks in an automotive assembly plant. Based on the field observations, the patterns of foot motions during object transfer tasks were hypothesized to be usefully described by a hybrid discrete/continuous model structure that predicts both strategies (behaviors) and the scaling of step placements within behaviors based on statistical analyses of laboratory data. A similar set of stepping behaviors was observed for both the laboratory and plant study. Contrary to expectations, approximately 70% of the object interactions (pickup or place operations) in both the plant and laboratory occurred with only one foot in contact with the ground. The four most frequently utilized strategies in the laboratory study accounted for 81% of the behaviors observed in the plant study. The results suggest that a majority of the stepping progressions used for MMH transfer tasks can be represented by a concise set of scalable behaviors. Based on the laboratory study, a Transition Stepping (TRANSIT) model was developed to predict foot placements for MMH tasks based on the task and operator characteristics. The model uses a new Quantitative Transition Classification System (Q-TRACS) that defines step placements during object pickups and deliveries. Multiple regression models were developed for five common stepping strategies. The placement and orientation of the terminal stance lead foot (i.e., the primary support foot) was predicted moderately well by subject and task attributes (R2 of 0.69, 0.43, and 0.68 for lateral placement, fore-aft placement, and foot orientation, respectively). Errors were typically less than one foot length. An Integrated Stepping Model (ISM) was developed that integrates the TRANSIT model with a model of gait using a flexible scaling structure to mediate between gait and transition stepping. The ISM uses the new concept of principal steps to smoothly join gait to transition stepping. The ISM is demonstrated using a range of work-cell scenarios with interspersed gait and acyclic stepping. The results of this research have direct application to ergonomic analysis of industrial tasks using digital human models, but also provide for the first time in the literature an integrated framework for representing both cyclic and acyclic stepping patterns with the quantitative detail required for biomechanical simulation and analysis.