Climate models are indispensable tools for quantitatively predicting changes in global mean surface temperature in response to a doubling of atmospheric carbon dioxide concentrations. Despite decades of effort within the modeling community, large uncertainties persist. This opinion paper advocates for prioritizing the use of perturbed parameter ensembles (PPEs) as an effective strategy for reducing uncertainty in climate projections, rather than focusing on increases in model complexity or spatial resolution. Based on a synthesis of the existing literature, the authors evaluate the broad applicability and demonstrated effectiveness of PPEs and propose a strategic agenda for their expanded use within the climate modelling community.