There is increasing scrutiny of pharmaceuticals on their value proposition as well as a growing demand for evidence on real world effectiveness once they are commercially available. There are many challenges in producing valid and reliable estimates of real world effectiveness. A major challenge is assessing a product’s effectiveness relative to why patients may respond differently to a treatment (i.e., identifying groups of patients exhibiting “Heterogeneity of Treatment Effect” (HTE) using subgroup identification methods). Assessing HTE is critical to understanding differences that may exist between the efficacy observed in randomized clinical trials and a product’s real world effectiveness. Understanding causes for HTE is required for correct attribution of any observed difference between efficacy and effectiveness to the product versus other sources (e.g., patient behavior); Not recognizing and accounting for HTE will confound assessment of a product’s performance, which ultimately affects its acceptance and use by payers, physicians, and patients. Failure to define and incorporate subgroups is a frequent criticism of systematic evidence reviews and comparative effectiveness research reports. However, the analytical methods for finding factors that define subgroups that explain HTE are challenging due to many known statistical issues (e.g., limited statistical power, multiplicity adjustments) Real world data exacerbates the analytical challenges due in part to biases (e.g., selection bias) and issues (e.g., data quality) inherent in the data. We will describe the data and bias challenges that create these analytical complexities for detecting the cause and magnitude of HTE when using real world data. We will present results from a simulation experiment that compared and validated several subgroup methods developed to address these data and analytical issues. We simulated 22 permutations of subgroups with known identification criteria and treatment effects to determine the performance of the methods .