Introduction: Few reliable measurements are available to help the medical staff in the subspecialty trainee selection process and the Letters of Recommendation (LORs) are an example.one of them. We intend to determine the most important LOR elements associated with increased odds of interviewing at our Institution and of matching in a Gastroenterology and Hepatology fellowship (GHF). Methods: The 2016 Mayo Clinic Arizona GHF candidates LORs were reviewed and we extracted data according to prior LOR publications and reviews: level of enthusiasm noted (eg: I have the honor of writing this LOR), noted waived reading rights, competencies (interpersonal skills, initiative, knowledge, leadership, professionalism, patient care) and if a numerical rank (eg: top 10% of all residents the author had worked with) was provided. Author data included: academic rank (Professor, Associate or Assistant), type of work (academic vs non academic), area of specialty (gi vs non gi), gender and if the author's institution ranked in the Top 20 Gastroenterology Hospitals (Best Hospitals survey 2016). Logistic regression was used in order to calculate the odds of interviewing or matching. The Hochberg step-up procedure for multiple testing was used to adjust the p values, which if less than 0.05 were considered statistically significant. Results: The LORs of 367 GHF candidates were reviewed. 1435 were written by academic authors (97%), 1013 written by gastroenterologists (68.4%), 1360 were men (91.9%) and 439 were professors (29.7%) (Table 1). A significantly higher likelihood of interviewing was found when the LORs came from a Top 20 Institution (OR 3.22, CI 2.01-5.02, p < 0.001). A significantly higher likelihood of matching occurred with LORs written by a gastroenterologist (OR 1.42, CI 1.12-1.77, p=0.04), from a Top 20 program (OR 1.80, CI 1.37-2.35, p=0.0007), if the letter mentioned the author's level of enthusiasm (OR 2.79, CI 1.92-4.05, p < 0.0001), if it noted the candidate's waived reading right (OR 2.14, CI 1.73-2.65, p < 0.0001), and if the letter reported the candidate's professionalism (OR 1.36, CI 1.32-2.01, p=0.0002) (Table 2 and 3).Table: Table. Descriptive StatisticsTable: Table. Odds of Interviewing and Odds of MatchingConclusion: This is the first study evaluating the use of LORs in the GHF application process. Certain LOR components and author characteristics predict a higher likelihood of interviewing and matching. Both applicants and academic staff should take into consideration these findings when requesting a LOR and when composing these letters, respectively.