Online gamblers are more likely to experience gambling problems. The main objective was to compare the frequency of recovery one (middle-term) and two (long-term) years after treatment initiation, according to the gambling medium (Internet vs. land-based gambling). The secondary objectives were (i) to compare online and offline gamblers at inclusion and (ii) to investigate whether the gambling medium was a predictive factor of recovery. Outpatients beginning treatment for a GD (n = 237) were assessed at inclusion (treatment initiation) and after 1 and 2 years. Bivariate analyses were performed to compare online and offline gamblers at inclusion and on the frequency of recovery at one and two years. Two multivariate logistic regressions were then performed to identify factors associated with middle- and long-term recovery. The majority of patients achieved middle (74.2%) and long-term (78.9%) recovery, with no difference between online and offline gamblers. The gambling medium was not a predictive factor of recovery. Patients with a higher perceived self-efficacy (OR = 1.04 [1.01-1.07], p = .046) and having no history of mood disorders (OR = 11.18 [2.53-49.50], p < .001) at inclusion were more likely to achieve middle-term recovery, while long-term recovery was associated with a lower level of sensation seeking (OR = 0.67 [0.48-0.92], p = .015) at treatment initiation. Online gambling did not seem to influence middle- and long-term recovery compared to offline gambling. Enhancement of perceived self-efficacy and treatment of mood disorders, and treatment strategies focused on sensation-seeking may represent helpful care strategies for favouring achievement of middle-term recovery and maintenance of long-term recovery, respectively. ClinicalTrials.gov NCT01248767, date of first registration: November 25, 2010.