Electronic Health Records (EHR)-derived real-world data (RWD) are large-scale observational databases of patient characteristics, treatment patterns, and outcomes. An emerging use case of RWD is generating real-world comparator cohorts (rwCC), where a curated dataset of real-world patients receiving standard of care may be used to contextualize the benefit of an investigational therapy studied in a single arm clinical trial. Construction of rwCCs includes aligning to target trial’s eligibility criteria and enrollment timing, balancing (e.g. matching or weighting) on key prognostic characteristics, and comparing real-world and clinical trial outcomes. In this talk, we discuss analytic considerations for constructing rwCCs, such as missing data and post-baseline variables (for example subsequent therapy received). We illustrate these concepts in a case study replicating the control arm of an oncology trial, using data from the Flatiron Health EHR-derived de-identified database.