The accurate prediction of welding distortion is an important requirement for the industry in order to allow the definition of robust process parameters without the need to perform expensive experiments. Many models have been developed in the past decades in order to improve prediction. Assumptions are made to make the models tractable; however, the consequences are rarely discussed. One example for such an assumption is the strain hardening model, which is often a choice between either kinematic or isotropic hardening. This paper presents the results of tensile tests for DP600 performed from room temperature up to one thousand degrees and for different strain-rates. In order to employ a mixed isotropic-kinematic hardening model, the fractions of each hardening contribution have been determined by means of bend testing. The welding distortion of a DP600 overlap joint has been simulated and it is shown that such a mixed-hardening model results in more accurate and reliable results.