Background. New persistent opioid use occurs in 3% to 14% of patients after elective surgery, but is poorly described after cardiothoracic surgery. We examined the association of prescription size with new persistent opioid use after cardiothoracic surgery. Methods. Opioid-naive Medicare patients undergoing cardiothoracic surgery between 2009 and 2015 were identified. Patients who filled an opioid prescription between 30 days before surgery and 14 days after discharge and with continuous Medicare enrollment 12 months before and 6 months after surgery were selected (n [ 24,549). New persistent use was defined as continued prescription fills 91 to 180 days after surgery. Prescription size was reported in oral morphine equivalents. Multivariable regression was performed for risk adjustment, and new persistent use rate was estimated. Results. Patient age was 71 ± 8 years, 9222 (38%) were female, and 20,898 (85%) were white. Overall new persistent use was 12.8% (3153 of 24,549), and declined yearly from 17% in 2009 to 7.1% in 2015 (P < .001). Prescription size, preoperative prescription fills, black race, gastrointestinal complications, disability status, open lung resection, dual eligibility (Medicare and Medicaid), drug and substance abuse, female sex, tobacco use, high comorbidity, pain disorders, longer hospital stay, and younger age were associated with new persistent use. Adjusted new persistent use was 19.6% (95% confidence interval, 18.7% to 20.4%) among patients prescribed more than 450 oral morphine equivalents, compared with 10.4% (95% confidence interval, 9.9% to 10.8%) among those prescribed 200 oral morphine equivalents or less (P < .001). Conclusions. Size and timing of perioperative opioid prescriptions were the strongest predictors of new persistent opioid use after cardiothoracic surgery. Modifiable risk factors such as prescription size should be targeted to reduce new persistent use.