PURPOSE: To describe the most effective approaches to achieving a real-world study data sample that is representative of the target population. DESCRIPTION: Considerable attention is paid to the design and analysis of outcomes research studies to address internal validity by minimizing bias and confounding.  However, too often, study sample populations are simply assumed to be representative of the study populations from which they are drawn, or are assessed for their representativeness only after the study has been conducted. Ideally, sample estimates should be as close as possible to their population value in order to make inferences about that study population. Practical implementation of measures to avoid selection bias and ensure a robust sampling procedure can be problematic.  Challenges include, willingness of sites and patients to participate in research (convenience sampling), and management of site and patient drop-out after the study has begun.  While many database studies and patient registries carry very large sample sizes and therefore begin to approximate the target population simply by means of sheer size, smaller studies may need to take steps, through stratified sampling and enrollment caps, to ensure that the study sample is reflective of the target population. These stratification variables may be at the site level (e.g., physician specialty, geography), the patient level (age, gender, ethnicity, disease duration) or both.  Temporal issues may also be problematic where studies performed in the past may not reflect rapid changes impacting today’s target population.  Following a brief overview of the design and analysis considerations, this presentation will focus on case examples, drawn from different organizations, of approaches to achieving a representative sample, highlighting some of the challenges intrinsic to real-world research.  Best practice recommendations will be provided to guide researchers on the most effective approaches, including the use of reference populations within specific countries.