South Australia has some of the highest magnetotelluric (MT) station coverage worldwide providing a unique opportunity to understand lithospheric architecture of the state and ultimately reduce the exploration risk for mineral occurrences. While 2D modelling of profile data has become standard, many areas are prone to 3D effects and are inadequately modelled using a 2D approach. As a result, data collected in a semi-regular array and modelled in 3D has become more common. However, as 3D inversion of MT datasets is computationally expensive it can be difficult to explore the impact of varying modelling parameters on the final 3D resistivity model. We applied parameterisation tests to a 3D MT inversion of the AusLAMP dataset in the northeastern part of South Australia to test various modelling parameters. The impact of incrementally including prior information to the starting or prior model (bathymetry), then increased conductivity beneath 410 km (10 ?m), and finally further increased conductivity (1 ?m) beneath 660 km were tested. The results highlight that it is still critical to test parameters controlling the model space. We show the critical parameters for this dataset to be the starting resistivity of the model and the model covariance.