Wheelchairs are the most widely used assistive device to aid activities of daily living (ADL) for disabled people. However, manual pushing of a wheelchair frequently leads to overuse of upper extremities causing shoulder pain and carpal tunnel syndrome. The purpose of this study was to test a novel method of estimating temporal parameters of wheelchair propulsion using inertial sensors. In this paper, normalized coordinate values of the vector defined on the upper arm were calculated from an inertial sensor attached on the upper arm. The number of strokes and push cycle timings including duration of propulsion and recovery phases were estimated for steady state wheelchair propulsion. This estimation was completed using a novel vector-based approach and a previously published resultant acceleration method; both were compared to timings measured using the SmartWheel. Measurements were performed on level and sloped surfaces with 10 able bodied subjects. The vector-based method improved estimation of the number of strokes when compared to the resultant acceleration method. However, the push cycle was estimated with better accuracy by the resultant acceleration method. Therefore, a combination of the vector-based and the resultant acceleration methods is proposed to ensure more accurate estimation of temporal parameters. The results suggest inertial sensors can be used to measure wheelchair activity accurately and reliably.