Gastrointestinal bleeding results in significant morbidity, mortality, and healthcare costs in the United States. The Center for Medicare and Medicaid Services' payment reform programs assess quality and value based on rates of hospital readmission for patients with gastrointestinal bleeding, but they identify these patients using Medicare Severity Diagnosis Related Groups (MS-DRGs), which include many types of gastrointestinal bleeding and do not account for the clinical heterogeneity among these patients. We aimed to characterize heterogeneity in outcomes of subgroups of patients with gastrointestinal bleeding. We performed was a cross-sectional, claims-based retrospective analysis of Medicare fee for service beneficiaries hospitalized for gastrointestinal bleeding in 2014 (159,000 hospitalizations). The primary outcome was unplanned readmission within 30 days of discharge from the hospital (30-day readmission). Secondary outcomes included length of stay, inpatient mortality, and death within 30 days of admission to the hospital (30-day mortality). Analyses were adjusted for age, sex, race, and Elixhauser comorbidities using logistic and Poisson regression, adjusting for clustering within hospitals. The 30-day readmission rate was 16.0%. Readmission rates varied among patients with different types of gastrointestinal bleeding, ranging from 13.5% for diverticular bleeding to 18.6% for small bowel bleeding. The mean length of stay was 4.2 days and 30-day mortality was 6.9% (ranging from 3.4% for diverticular bleeding to 12.1% for upper gastrointestinal bleeding not otherwise specified). When hospitalizations were stratified by MS-DRGs, the main source of variation in rates of readmission and mortality was MS-DRGs. In a retrospective analysis of Medicare fee for service beneficiaries hospitalized for gastrointestinal bleeding, we found that 16% of these patients are readmitted to the hospital. Rates of hospital readmission, length of stay, and mortality vary with type of gastrointestinal bleeding, but MS-DRGs account for the largest source of variation. Policies focused on quality and value should account for this heterogeneity.