Background: Oral budesonide has been proven effective in short- and long-term treatment of collagenous colitis; however, symptom relapse frequently occurs after drug withdrawal. The aim of this study was to identify the risk factors for symptom relapse in patients with collagenous colitis after withdrawal of short-term budesonide therapy. Methods: One hundred twenty-three patients from 4 randomized controlled studies who achieved clinical remission after short-term treatment with budesonide (9 mg/d) were analyzed, including 40 patients receiving subsequent budesonide maintenance therapy (6 mg/d) for 6 months and 83 patients without active maintenance treatment. Variables available for analysis were age, sex, baseline stool frequency, duration of diarrhea, collagenous band thickness, and lamina propria inflammation. Hazard ratios (HRs) and their 95% confidence intervals (CIs) were calculated by Cox proportional hazard model. Results: The overall symptom relapse rate was 61%. By multivariate analysis, a baseline stool frequency >5 per day (HR, 3.95; 95% CI, 1.08-14.39), history of diarrhea >12 months (HR, 1.77; 95% CI, 1.04-3.03), and the absence of budesonide maintenance therapy (HR, 2.71; 95% CI, 1.37-5.38) were associated with symptom relapse. The time to relapse was shorter in patients with a baseline stool frequency >5 per day (56 versus 199 d, P = 0.024), as in those with history of diarrhea >12 months (56 versus 220 d, P = 0.009). Budesonide maintenance therapy delayed the time to relapse (56 versus 207 d, P = 0.005). Conclusions: Our data demonstrate that a high stool frequency at baseline and a long duration of diarrhea are risk factors for symptom relapse in collagenous colitis, whereas budesonide maintenance therapy is a protective factor against symptom relapse.