Using several different methodologies, we quantify the statistical robustness of variables used in prior research to explain initial IPO returns. We establish a parsimonious list of robust variables and evaluate their implications for different theories of IPO underpricing and clustering. Further, we illustrate how using such a set of robust explanatory variables leads to several different conclusions than prior research that failed to include these important control variables. Researchers who identify new potential predictors of IPO initial returns should control for the list of robust variables we identify.