For people with physical disabilities, manual wheelchairs are essential enablers of mobility, participation in society, and a healthy lifestyle. Their most general design offers great flexibility and direct feedback, but has been described to be inefficient and demands good coordination of the upper extremities while critically influencing users’ actions. Multiple research groups have used Inertial Measurement Units (IMUs) to quantify physical activities in wheelchairs arguing that knowledge over behavioural patterns in manual wheelchair usage can guide technological development and improved designs. The present study investigates turning behaviour among fulltime wheelchair users, laying the foundation of the development of novel steering systems that allow directing kinetic energy by means other than braking. Three wearable sensors were installed on the wheelchairs of 14 individuals for tracking movement over an entire week. During detected “moving windows”, phases where the velocities of the two rear wheels differed by more than 0.05 m/s were considered as turns. Kinematic characteristics for both turns-on-the-spot as well as for moving turns were then derived from the previously reconstructed wheeled path. For the grand total of 334 km of recorded wheelchair movement, a turn was detected every 3.6 m, which equates to about 900 turns per day on average and shows that changing and adjusting direction is fundamental in wheelchair practice. For moving turns, a median turning radius of 1.09 m and a median turning angle of 39° were found. With a median of 89°, typical turning angles were considerably larger for turns-on-the-spot, which accounted for roughly a quarter of the recognised turns and often started from a standstill. These results suggest that a frequent pattern in daily wheelchair usage is to initiate movement with an orienting turn-on-the-spot, and cover distances with short, straightforward sections while adjusting direction in small and tight moving turns. As large bends often require simultaneous pushing and breaking, this is, perhaps, the result of users intuitively optimising energy efficiency, but more research is needed to understand how the design of the assistive devices implicitly directs users’ movement.