Current simulations of the injection molding process insufficiently consider the thermal interactions between melt, solidified material and mold. To improve the prediction accuracy in terms of shrinkage and warpage these interactions cannot be neglected and require a precise observation of the temperature field within the ongoing process. Nowadays temperature measurements at the transition of the polymer melt to the mold or near the surface of the polymer melt are applicable and do not allow either a non-invasive analysis or a determination of the temperature field. In the approach of ultrasound tomography presented here, an ultrasound beam is emitted into the melt and the time-of-flight (TOF) is detected by a set of transducers, which are radially arranged around the melt. Subsequent the measurement is repeated from different directions. Using algebraic reconstruction techniques, a distribution of the ultrasound velocity can be calculated based on the TOF-dataset. With additional information about the polymer from pvT-curves, the distribution of the ultrasound velocity can be converted into a temperature field. To evaluate the applicability of ultrasound tomography, a prototypical measurement device was set up. The device consists of a hollow cylinder in which eight brass forerun elements are integrated in order to scatter the ultrasound beam. A polymer melt can be placed within the hollow cylinder and ultrasound tomography can be performed. The results show the applicability of commercially available transducers and the operability of the brass forerun elements, since each of the seven transducers was able to detect a signal that was sent out from the one remaining transducer. However, a useful reconstruction of the temperature distribution is, due to the inhomogeneous melt in the prototype, not yet possible. Currently an injection mold is designed which fulfills all prerequisites for ultrasound tomography and overcomes the difficulties from the prototype.