BACKGROUND: The management of intra-abdominal fistulizing Crohn's disease involves surgical resection and biologic therapy. The criteria for choosing one therapy over the other are debated. OBJECTIVES: Identify factors influencing the choice of early surgical intervention over biologic therapy. DESIGN: Retrospective. SETTING: Single center, tertiary training and research hospital. PATIENTS AND METHODS: We analyzed adult patients with Crohn's disease who had intra-abdominal fistulas and were followed for three years. Baseline data were collected from medical records, and imaging studies assessed the fistula type, number, affected segment length, and presence of strictures and abscesses. Multivariable logistic regression analysis was used to identify predictors for surgical intervention. MAIN OUTCOME MEASURES: Factors that led to early surgical intervention in patients with intra-abdominal fistulizing Crohn's disease. SAMPLE SIZE: 73 patients. RESULTS: Seventy-three patients met the inclusion criteria: 27 (37.0%) in the nonsurgical group and 46 (63.0%) in the surgical group. Early surgical intervention was done if patients had bloating or constipation (P=.018), extensive disease segments (P<.001), and no prior biologic treatment (0.015). In the multivariate analysis, early surgical intervention was indicated for enterocutaneous fistulas (odds ratio [OR]: 8.20, 95% confidence interval [CI]: 1.25-53.80, P=.03), abscesses (OR: 5.18, 95% CI: 1.03-26.12, P=.046), and strictures (OR: 6.08, 95% CI: 1.26-29.25, P=.024). Nonsurgical fistula treatment resulted in complications in 55% of patients, 48% of them requiring surgical resections, whereas biologic treatment achieved a 40.7% fistula healing rate. CONCLUSIONS: Findings associated with Crohn's fistulas, including enterocutaneous fistulas, extensive disease segments, strictures, and abscesses, are associated with a higher likelihood of early surgical intervention and may suggest potential ineffectiveness of biologic therapies. LIMITATIONS: This was a retrospective analysis of a single center with a small sample size, which may involve a degree of recall bias when data are collected, thus reducing the reliability of the results.