In this work three different cross section groups of stainless steel T-Spring, for tooth retraction, have been tested; each spring is activated for 1 mm, 2 mm, and 3 mm, and the resultant force system is evaluated by using a testing apparatus. The results showed that when the cross section and activation distances are increased, the horizontal force and moment increased, while for the moment-to-force ratio, the lowest mean value was at the first activation distance of the first group, and the highest mean values were at the third activation distance of the third group. All three groups at all activation distance are insufficient to produce bodily tooth movement. T-springs of the cross section and with frequent activation provide the best in force system production. An artificial neural network model was trained for simulation of the correlation between input parameters: spring cross section and activation distance, and the outputs spring force system. The network model has prediction ability with low mean error of force prediction (5.707%), and for the moment is (4.048%), and it can successfully reflect the results that were obtained experimentally with less costs and efforts.