Objective To determine the long-term impact of extravasation at initial post-urethroplasty urethrogram, we present a novel classification system and report the association of leak severity with outcome. Materials and Methods A total of 91 patients underwent buccal graft urethroplasties of the bulbar urethra from 2007 to 2015. Median time to urethrogram was 23 days. All leaks were characterized by length and width. Cutoffs for length and width were calculated using receiver operating characteristic curves. Each urethrogram was graded 0-3 (0 = no leak) with 1 point given for any leak, length ≥1.03 cm, and width ≥0.32 cm. Failure was stricture recurrence on cystoscopy. One-year failures were compared using Fisher test. Kaplan-Meier curves were constructed to measure the impact of variables on recurrence. Results Mean age was 46.1 years. Of 91 urethroplasties, 31 had extravasation on initial imaging. With median follow-up of 11 months, 15 patients had stricture recurrence, 6 of whom had leak on initial urethrogram. For patients with at least 1 year of follow-up, there was no difference at 1 year for failures, as to any leak (P = .220), length (P = 1.000), width (P = 1.000), or grade (P = .823). Grade 3 was associated with higher failure rates compared with grades ≤2 using Kaplan-Meier curves (P = .031), with the curves significantly diverging around the 1-year mark. Similarly, length ≥1.03 cm was associated with higher failure rates compared with <1.03 cm beyond 1 year (P = .044). Conclusion Although there is no indication that leaks at 3 weeks are associated with short-term failure, leak length and width appear to be a predictor of longer term recurrence (>1 year).