Incremental Dynamic Analysis (IDA) is a powerful method for evaluating the seismic performance of structural systems. When undertaking IDA, ground motion records are scaled in accordance with a certain seismic intensity level up until the limit of collapse of the structure is reached. Peak ground acceleration (PGA) is often used as intensity measure (IM) for scaling purposes whilst peak ground velocity and response spectral accelerations can also be used as IM parameters. One of the challenges with the IDA methodology is that the performance (IDA) curve obtained for a structural system is non-unique meaning that it can be sensitive to changes in the frequency properties of the input excitations. Thus, there is a practice of holding the frequency content of the excitation unchanged (or constrained to a model design spectrum) when applying the scaling. The shortcomings with this modelling approach is that real conditions can be misrepresented given that frequency contents of the earthquake changes with magnitude and distance. This paper presents findings from an investigation which was aimed at determining what ground motion parameters are most suitable for use as IM whilst allowing the frequency content of the earthquake to vary in an IDA. This paper presents findings of an investigation wherein nonlinear dynamic time-history analyses involving the use of both recorded and artificial accelerograms were undertaken on models representing buildings constructed of cold-formed steel. For systems with natural period of less than 0.4s; PGA and the peak acceleration demand (PAD) have been found to be desirable choices of IM parameters given that the associated IDA curves are generally insensitive to the choice of the accelerogram ensemble used. For systems possessing longer natural periods of vibration, other ground motion parameters have been found to be more desirable choices.