Earth System Models (ESMs) are heavily used as inputs to human-relevant impact models and multisectoral dynamic models. Therefore, representing the full range of model uncertainty, scenario uncertainty, and interannual variability displayed by ESMs participating in CMIP6, in their role as inputs to other models, is critical to understanding the future co-evolution of the integrated human-Earth system. However, the increased participation by modeling centers in CMIP6 relative to previous eras has resulted in a large archive of ESM data that can be intractable for impacts modelers to effectively utilize, due to computational constraints. This is also likely to continue to grow moving into future eras of CMIP and other major model intercomparisons. We will present a hybrid quantitative-heuristic method to select a subset of CMIP6 models for use as inputs to a impacts or multisectoral model, while still representing the range of model uncertainty, scenario uncertainty, and internal variability of the full CMIP6 ESM results. This method is intended to help human-relevant impact modelers select climate information from the CMIP archive. Our examples are tailored to focus on temperature and precipitation outputs of ESMs, as these are two of the most used variables among impacts models and many other key input variables for impacts are at least correlated with one or both of temperature and precipitation (e.g. relative humidity). We also prioritize selecting ESMs in the subset with multiple ensemble members available for scenarios, where possible. This approach could be applied to other output variables of ESMs and, when combined with emulators, offers a flexible framework for designing more efficient experiments on human-relevant climate impacts.