Testing of steel roof purlins is usually performed on full scale models in large vacuum test rigs. To undertake a comparison between web cleat connected purlins and flange bolted purlins a series of tests were performed on a 1:4 small scale model vacuum test rig. Various modelling issues need to be addressed to ensure reasonable comparison with actual constructed roof framing methods but still be suitable for an economical comparison between the connection methods. Model test results were supported by, and found to be in reasonable agreement with, deflection predictions from computer models based on finite element methods. This paper discusses the testing methods adopted and the value of small scale model testing programs as a means of obtaining comparisons between framing options.