Multipass butt welding of a very thick steel plate has been performed. Transient temperatures and residual stresses have been measured. The agreement between calculations and experiments is good. Two different approaches, quiet and inactive elements, for modelling multipass welding are compared. The first approach is straightforward to apply in most finite element codes. The inactive element method requires a code that can regenerate the finite element model automatically or otherwise very tedious manual work is necessary as the elements are added to the model when welds are laid. It is shown that both techniques give the same results but the computational effort is reduced by using inactive elements. It also circumvents the problem in the quiet element approach of choosing properties of elements in the model that represent the case when welds are not laid.