The objective of this study was to characterize the changes in adverse events, seizure severity, and frequency in response to a pharmaceutical formulation of highly purified cannabidiol (CBD; Epidiolex®) in a large, prospective, single-center, open-label study. We initiated CBD in 72 children and 60 adults with treatment-resistant epilepsy (TRE) at 5 mg/kg/day and titrated it up to a maximum dosage of 50 mg/kg/day. At each visit, we monitored treatment adverse events with the adverse events profile (AEP), seizure severity using the Chalfont Seizure Severity Scale (CSSS), and seizure frequency (SF) using seizure calendars. We analyzed data for the enrollment and visits at 12, 24, and 48 weeks. We recorded AEP, CSSS, and SF at each follow-up visit for the weeks preceding the visit (seizures were averaged over 2-week periods). Of the 139 study participants in this ongoing study, at the time of analysis, 132 had 12-week, 88 had 24-week, and 61 had 48-week data. Study retention was 77% at one year. There were no significant differences between participants who contributed all 4 data points and those who contributed 2 or 3 data points in baseline demographic and AEP/SF/CSSS measures. For all participants, AEP decreased between CBD initiation and the 12-week visit (40.8 vs. 33.2; p < 0.0001) with stable AEP scores thereafter (all p ≥ 0.14). Chalfont Seizure Severity Scale scores were 80.7 at baseline, decreasing to 39.2 at 12 weeks (p < 0.0001) and stable CSSS thereafter (all p ≥ 0.19). Bi-weekly SF decreased from a mean of 144.4 at entry to 52.2 at 12 weeks (p = 0.01) and remained stable thereafter (all p ≥ 0.65). Analyses of the pediatric and adult subgroups revealed similar patterns. Most patients were treated with dosages of CBD between 20 and 30 mg/kg/day. For the first time, this prospective, open-label safety study of CBD in TRE provides evidence for significant improvements in AEP, CSSS, and SF at 12 weeks that are sustained over the 48-week duration of treatment.