Spring back is the major issue in the bending of tubes. Presently trial and error methods are followed in the production of components for compensating the spring back problem. This takes an ample amount of time and money for the industries. More quantity of components gets rejected due to this problem. So this makes the companies to go for a solution to this problem. In this project the spring back analysis of hollow cylindrical tubes of various thickness and diameter is made. In this project we propose artificial neural network in which initial parameters are fixed to get desired final shape. An experimental method of bending the tubes is done to obtain the initial parameters. Then the initial parameters comprising of input values and target values are taught to the artificial neural network in the learning cycles. The network error is reduced by doing further iterations. Then the validation of the values is made. Now the input value is given to get the required output value. Comparison between the results obtained by both artificial neural network and experimental method is carried out to show the integrity of artificial neural network. Now we can get to compare the required spring back value from artificial neural network and experimental method within the area taught to it.