Background and aims: Opioids are commonly prescribed to patients with chronic low back pain (LBP) despite risks of harms. We conducted a discrete choice experiment (DCE) to determine factors contributing to a general practitioner’s (GP’s) decision to prescribe either an opioid or an NSAID to a patient with chronic LBP. Methods: GPs recruited through an online survey distributed in Australia were presented with 12 questions that represented hypothetical clinical scenarios of a patient with chronic LBP. The clinical scenario varied by two patient attributes; LBP with or without referred leg pain (sciatica) and comorbidities. Participants chose their preferred alternative either an opioid, NSAID or neither (“opt-out”). Each alternative varied by three clinical attributes: the type of opioid or NSAID, the degree of pain reduction and number of adverse events. Results: 210 GPs participated in the survey. Overall, GPs preferred to prescribe an NSAID (45.2%, 95% CI 38.7% to 51.7%) over an opioid (28.8%, 95% CI 23.0% to 34.7%). However, there was no difference between the type of NSAID or opioid preferred. Patient attributes of comorbidities (zero, one, two or three), and the presence of referred leg pain (sciatica) did not influence prescribing preferences, nor did clinical attributes of pain reduction and adverse events. Conclusions: GPs prefer to prescribe an NSAID over an opioid for a patient with chronic LBP. This preference appeared fixed and was not changed by clinical (drug type, degree of pain reduction or number of adverse events) or patient attributes (comorbidities or presence of referred leg pain).