Steel design codes continue to be expanded to permit the use of more advanced methods of non-linear analysis. Designers looking to employ such methods need to validate their analysis software and, just as importantly, verify their ability to utilize it properly. The literature contains many benchmark problems and results to help achieve this, but nearly all are limited to two-dimensional behaviour. This paper is intended to contribute a new set of benchmark problems in order to help satisfy the need for a database of examples in which accurate modelling of three-dimensional or spatial behaviour is essential.