OBJECTIVES:  Little is known about usual care treatment patterns and associated outcomes in Multicentric Castleman Disease (MCD). Information on the management of this disease can inform clinical practice, treatment guidelines, and elucidate areas of unmet medical need.  The design and execution of a retrospective chart review study of patients with MCD are described, highlighting general methodological considerations for conducting chart review studies in rare diseases.   METHODS:  A multi-center, retrospective, chart review study of 59 MCD patients (61.0% male; mean age 53.3 ± 16.3 years) was conducted in two centers in the United States.  All MCD cases within a defined eligibility period were identified. For eligible patients, medical record data were abstracted by site study staff; up to six months pre-index diagnosis date and up to three years of post-diagnosis. Anonymized data were recorded on paper case report forms (CRF) and entered into an electronic data capture (EDC) system. Key design challenges and lessons learned include: 1) site recruitment: limited number of participating treatment centers resulting in small study population; 2) CRF design: disease complexity and lack of published literature necessitated the involvement of MCD clinical experts with management knowledge 3) data abstraction: patients only seen at the site for a one-time consultation or second opinion resulting in minimal and missing data, making MCD diagnosis confirmation and eligibility criteria difficult to confirm and; 4) EDC system design: data from multiple clinical tests, exams and physician visits were collected, therefore, the EDC system must permit large volumes of data with a validation plan to ensure quality data.   CONCLUSIONS:  Acquisition of clinical outcomes, resource utilization and treatment pattern data through retrospective chart review study in a rare disease population such as MCD is challenging and requires an innovative approach, disease knowledge, and sufficient treatment centers to optimize external validity.