Most manual wheelchair users will experience upper extremity injury and pain during their lifetime, which can be partly attributed to the high load requirements, repetitive motions and extreme joint postures required during wheelchair propulsion. Recent efforts have attempted to determine how different propulsion techniques influence upper extremity demand using broad measures of demand (e.g., metabolic cost). However studies using more specific measures (e.g., muscle stress), have greater potential to determine how altering propulsion technique influences demand. The goal of this research was to use a musculoskeletal model with forward dynamics simulations of wheelchair propulsion to determine how altering propulsion technique influences muscle demand. Three studies were performed to achieve this goal. In the first study, a wheelchair propulsion simulation was used with a segment power analysis to identify muscle functional roles. The analysis showed that muscles contributed to either the push (i.e. delivering handrim power) or recovery (i.e. repositioning the hand) subtasks, with the transition period between the subtasks requiring high muscle co-contraction. The high co-contraction suggests that future studies focused on altering transition period biomechanics may have the greatest potential to reduce upper extremity demand. The second study investigated how changing the fraction effective force (i.e. the ratio of the tangential to total handrim force, FEF) influenced muscle demand. Simulations maximizing and minimizing FEF both had higher muscle work and stress relative to the nominal simulation. Therefore, the optimal FEF value appears to balance increasing FEF with minimizing upper extremity demand and care should be taken when using FEF to reduce demand. In the third study, simulations of biofeedback trials were used to determine the influence of cadence, push angle and peak handrim force on muscle demand. Although minimizing peak force had the lowest total muscle stress, individual stresses of many muscles were >20% and the simulation had the highest cadence, suggesting that this variable may not reduce demand. Instead minimizing cadence may be most effective, which had the lowest total muscle work and slowest cadence. These results have important implications for designing effective rehabilitation strategies that can reduce upper extremity injury and pain among manual wheelchair users. / text