A new geometry of orthodontic retraction spring was experimentally studied through an electronic device (platform for measuring forces), using strain gauges that were adapted to cantilever beams. The sample consisted of 36 titanium-molybdenum springs, divided into three groups of 12 springs each. The springs were produced with different cross sections of 0.016 x 0.022 inch and 0.017 x 0.025 inch and with different angles between the extremities (120 degrees and 130 degrees). The springs were adapted to the platform in three different positions so that the force system developed by them could be known (horizontal forces, vertical forces, alpha-beta moments, and moment-to-force ratio M:F). The analysis of factorial variance and the Tukey honestly significant difference test were applied to verify the differences between the averages caused by three possible variation sources and the interactions between them. Regression analysis was also performed to obtain the spring rate. The results show the interactions between the three geometric variables, force magnitudes, and also the spring rates, which are compatible with the ones mentioned in the literature related to the subject. The spring rate was within the levels that are appropriate for clinical use (varying from beta = 33.1 gf/mm to beta = 43.9 gf/mm).