Background: In response to the opioid crisis, recent policies aiming to reduce opioid prescribing, misuse, & abuse have generated concern that patients with cancer pain may unintentionally experience reduced access to necessary opioid therapy. It is unknown how opioid prescribing patterns have changed between generalists and oncologists during this era. Methods: We conducted a longitudinal repeated cross-sectional study estimating adjusted annual national trends in opioid prescribing among generalists & oncologists using the Medicare Part D Prescriber Public Use Files 2013-2017. Poisson models estimated annual adjusted predicted mean rates of opioid prescribing-per-1,000 total prescriptions & long-acting opioid prescribing per-1,000 opioid prescriptions. Poisson models estimated adjusted incidence rate ratios (aIRRs) to quantify annual changes in prescribing rates. Results: From 2013-2017 the annual adjusted predicted mean rate of opioid prescriptions per 1,000 total prescriptions decreased from 53.4 to 41.3 among generalists (aIRR = 0.78; p < 0.01) and from 133.2 to 105.9 among oncologists (aIRR = 0.83; p < 0.01). The rate of long-acting opioid fills per 1,000 opioid prescriptions decreased from 96.0 to 87.0 (aIRR = 0.87; p < 0.01) and 235.1 to 222.5 (aIRR = 0.95; p < 0.01) for generalists & oncologists, respectively (Table). Conclusions: We found large declines in overall opioid prescribing rates among generalists (-22%) and oncologists (-17%) from 2013-2017. Long-acting opioid prescribing rates decreased over 2.5-times more among generalists than oncologists. Opioid policy & advocacy have been effective in reducing the extent of opioid prescribing in the Medicare population but how much of the decrease in prescribing by oncologists is ‘appropriate’ versus ‘inappropriate’ deserves further investigation.