The anatomical model propulsion system (AMPS) was designed to propel manual wheelchairs in a highly repeatable manner while emulating human weight distribution and force application. This paper details the development of two control systems for the AMPS and their application on several sets of experiments. The first system allows the AMPS to maneuver with precision along straight and curvilinear trajectories. The second system mimics human pulsatile propulsion of a manual wheelchair for tests on a dynamometer. In both cases, the controller used an estimation of input forces provided by a model along with real-time feedback to create an appropriate maneuver control of the wheelchair. Several studies have measured rolling resistance using diverse methodologies and equipment including dynamometers, treadmills, and instrumented wheelchairs. A new approach for testing rolling resistance by using the AMPS is presented in this work. This new approach was able to test the wheelchair behavior on actual floor while all wheels are making contact, with precise control on velocity and acceleration, a combination unattainable with previous methods. Results confirm other researchers' conclusion that rolling resistance increases with velocity, but also add new evidence that rolling resistance increases significantly with acceleration on a manual wheelchair. The AMPS was also used to estimate turning resistance in manual wheelchairs. Results offer new evidence that turning resistance increases as the instantaneous radius of rotation of the trajectory shortens. Additionally, the AMPS was provided with the ability to propel a manual wheelchair emulating human pulsatile propulsion on a single-roller dynamometer. Various pushing frequencies and duration of pulses were tested to compare the efficiency of different propulsion techniques. Two indices, energy conversion efficiency (g) and cost of transport (COT), were used to quantify wheelchair mechanical efficiency due to their relevance for vehicles. Energy conversion efficiency was found to vary significantly for different values of acceleration. COT was found to increase as linear velocity increased and the radius of curvature was reduced. Initial findings on COT for these pulsatile propulsion experiments suggest propelling techniques might have an effect in overall mechanical efficiency. Wheelchair mechanical efficiency could be used to compare the performance of different wheelchair models over common maneuvers, helping clinicians do more informed decisions for their patients. Further development and exhaustive testing of the system presented in this work could potentially become a standard for the manual wheelchair industry.