The spring-tine harrow is gaining popularity for mechanical weeding. However, its weeding performance and mechanism have not been well understood. A spring-tine harrow was first tested in a controlled indoor soil bin at four different travel speeds (4, 6, 8, and 10 km h−1) with three different spring-loading settings (low, medium, and high). Then the harrow was tested in a wheat (Triticum aestivum L.) field at the same spring-loading settings at three different weeding timings (early, middle, and late) in 2019 and 2020. Soil cutting forces (draft and vertical), soil displacements (forward and lateral), soil working depth, weed control efficacy, weed density, and crop damage were measured. The results showed that the spring-loading setting had a more dominant effect on working depth and soil cutting forces than the speed. The soil displacements were more dependent on the speed compared with the spring-loading setting. Treatment effects on weeding performance indicators in the field were similar across years. Adjusting the spring-loading setting from low to high improved the weeding efficacy from 44.9% to 73.9% in 2019 and from 51.6% to 78.1% in 2020. Consequently, the final weed density was minimized at the high loading setting, with the reduction in 2020 being significant. The middle weeding timing caused the least crop damage, while reducing the final weed density by approximately one-third compared with the control (without mechanical weeding), which was the most desired outcome among the three timings tested.