Since one of the most important parameter in polymer processing such as injection stretch blow molding is temperature distribution in the thickness direction, an inverse method has been applied to estimate this profile. This process comprises of four steps. In the first step the preform is injection molded, and in the second and third step it is stretched by a rod to its final length and then inflated and in the last step it is discharged from the mold. In such kind of polymer flows viscous dissipation plays a remarkable role in the evolution of temperature profile. Some theoretical temperature profile has been applied to confirm the validation of the inverse algorithm. Different solution techniques are applied in this article to the inverse problem under consideration, namely: the conjugate gradient and Levenberg-Marquardt method. After the preform is injection molded, which is the first step, it is removed from the mold, which corresponds to time t = 0. At this moment an infrared camera is used to record the surface temperature of the preform with a certain time step. With regard to variation of thermal properties with temperature, the inverse problem becomes nonlinear. These experimental data provided by the infrared camera are then used to estimate the temperature profile at the end of injection process before stretching and inflation took place.