Building performance models routinely involve tens or hundreds of components or aspects and at least as many parameters to describe them. This results in overwhelming complexity and a tedious process if the designer attempts to perform parametric analysis in an attempt to optimize the design. Traditionally, during design, parameters are selected on a one-at-a-time basis and, occasionally, formal mathematical optimization is applied. However, many subsets of parameters show some level of interaction, to varying degrees, suggesting that the designer should consider manipulating multiple design parameters simultaneously. This paper is divided into two parts. The first part presents a methodology for identifying the critical parameters and two-way parameter interactions. The second part uses these results to identify the appropriate level of modeling resolution. The methodology is applied to a generic model for net-zero or near-net-zero energy houses, which will be used for an early stage design tool. The results show that performance is particularly sensitive to internal gains, window sizes, and temperature setpoints, and they indicate the points at which adding insulation to various surfaces has minimal impact on performance. The most significant parameter interactions are those between major geometrical parameters and operating conditions. Increased modeling resolution for infiltration and building-integrated photovoltaics (BIPV) only provides a modest improvement to simpler models. However, explicit modeling of windows, rather than grouping them into an equivalent area, has a significant impact on predicted performance. This suggests that identifying and implementing the appropriate level of modeling resolution is necessary, and that it should be detailed for some aspects even in the early stage design.