We study the impact of an information intervention on opioid prescribing using a preregistered research design and comprehensive nationwide register data. The intervention involved a personal letter sent to all Finnish physicians who had prescribed oxycodone or fentanyl to a patient who had purchased at least three months' supply of these medications in the previous year. These physicians were randomized into the treatment and control groups. The letter was sent to physicians in the treatment group in May 2019, and the control group received the same letter six months later. The intervention letter contained information about opioid use and proper pain treatment using opioids based on national clinical guidelines. While the intervention showed no significant effects in the whole study population, we detected heterogeneity in effect with respect to preregistered physician characteristics. We observed a 22% reduction in fentanyl and oxycodone prescriptions to new patients among physicians receiving their first information letter, a 4.8% reduction in any opioid prescriptions among high-volume prescribers as well as an increase of 7% in nonopioid analgesic prescribing among low-volume prescribers. These results highlight the challenges policymakers encounter when attempting to sustainably reduce opioid prescriptions and mitigate harmful clinical practices through repeated information-based interventions.