OBJECTIVE The objective of this study was to determine whether risk of early leukoplakia recurrence (within 3 months) following carbon dioxide (CO2) laser removal varies by clinical characteristics including lesion size, site and accessibility of margins. MATERIALS AND METHODS A retrospective cohort study included patients with oral leukoplakia who had their first CO2 laser surgery for removal of oral leukoplakia between 2005 and 2010 at the UCSF oral medicine clinic. Twenty-six patients with 32 separate lesions met the eligibility criteria after a clinic database search was followed by review of clinical notes and biopsy reports from existing patient charts. Data analysis included computation of summary statistics, and logistic regression analyses to evaluate recurrence of leukoplakia by clinical characteristics of the lesions. RESULTS Patient data and the characteristics of lesions were evaluated as possible predictors of early recurrence following laser removal; these included age, sex, duration, size, appearance and histopathology of the lesion. The only one that reached statistical significance was poor accessibility of the margins of the lesion (vs. good accessibility, OR = 24.57 (95% CI: 1.59-16.68), p = 0.016); the probability for trend for good, questionable, and poor accessibility was 0.0028. This finding remained significant after controlling for age, sex, duration and size of lesion. Four out of five lesions with poor accessibility showed recurrence at 3 months. Of these, three involved the gingiva and one the lateral tongue. CONCLUSIONS This study has identified poor accessibility of the lesion margins as a predictor for early recurrence of leukoplakia following laser removal. Other variables evaluated did not reach statistical significance, possibly due to lack of power.