Abstract Introduction: Granular cancer patient treatment data collection, and subsequent mapping to standard regimen definitions, are vital next steps in advancement of observational studies in oncology. However, the identification of regimen details, including dose and schedule, is a prerequisite for both collection and mapping. At the patient level, claims databases are a useful but limited resource. Most cancer registries, such as the National Cancer Institute Surveillance, Epidemiology, and End Results (SEER) program and the Commission on Cancer National Cancer Database, capture only a simplification of actual treatments, such as binary exposure to chemotherapy (yes/no) or a list of individual chemotherapeutic agent names. Moreover, even with optimal patient data collection, a “gold standard” database of expected chemotherapy doses and schedules as part of standard-of-care (SOC) does not currently exist. Methods: We extracted and normalized the semistructured dosing and timing data from HemOnc.org, the largest publicly available website of SOC chemotherapy drugs and regimens. To that end, we undertook two broad and parallel approaches: 1) standardization of prescribing instructions within a given cycle (SIGs) or pertaining to cycle timing (“cycleSIGs”); and 2) parsing of resultant content into structured variables. This effort was carried out iteratively with the goal of creating standard “canonical” forms for intermittent intravenous (IV), continuous IV (CIV), other routes (e.g., oral), and radiation SIGs. All SIGs were bound to regimen and treatment context (e.g., cyclophosphamide dosing differs in R-CHOP versus R-CVP, and number of cycles often differs between adjuvant and metastatic contexts). Results: There are currently 14,569 regimen-context-SIG-cycleSIG quartets in the database (October 2019). Parsing of SIGs into structured variables resulted in 7,792 canonical IV, 675 canonical CIV, 3,762 canonical other, 510 canonical radiation, and 2,249 noncanonical results. Some SIGs are multipart and were broken into steps. For example, “6 mg/m2 IV once on day 1, then 3 mg/m2 IV once on day 8” constitutes two separate steps. There were 948 unique cycleSIGs (e.g., “21-day cycle for 4 cycles”), which were also parsed into components. Discussion: This effort has produced a large dataset of granular drug and cycle SIG information that reflects SOC dosing parameters in hematology/oncology. This dataset can be used to understand discrepancies between real-world outcomes and clinical trial results, e.g., by elucidating the effect of dose reductions and treatment delays on treatment outcomes. The Observational Health Data Sciences and Informatics (OHDSI) oncology workgroup is arranging to add this new information to the portion of the HemOnc vocabulary that is available through the OHDSI terminology management tool. Ongoing efforts also include translating the maximum possible number of noncanonical sigs into canonical forms, which can further enhance simplicity and usability of the dataset. Citation Format: Zachary H. Moldwin, Andrew Malty, Donna R. Rivera, Anastasios Siapos, Christian G. Reich, Michael J. Gurley, Rimma Belenkaya, Dmitry Dymshyts, Jeremy L. Warner. Getting granular: A structured database of doses and schedules in hematology/oncology [abstract]. In: Proceedings of the AACR Special Conference on Advancing Precision Medicine Drug Development: Incorporation of Real-World Data and Other Novel Strategies; Jan 9-12, 2020; San Diego, CA. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(12_Suppl_1):Abstract nr 27.