OBJECTIVES: Treating chronic pain involves difficult trade-offs between the goals of pain relief and avoidance of side-effects, but evidence is scarce on relative perspectives of physicians and pain sufferers. We studied each group’s preferences concerning classical strong opioids in France, Germany, Italy, Spain, Sweden and the UK.  METHODS: In online, discrete choice experiments (DCE), chronic pain sufferers (n=242) and physicians (n=270) chose between hypothetical profiles or an opt-out in 15 choice tasks. Profile descriptions were based on attributes elicited in focus groups with 84 sufferers and semi-structured interviews with 11 physicians. Models were specified by multinomial logit and individual respondent part-worths were estimated by hierarchical Bayesian regression. RESULTS: All main-effects, but no interactions, were significant. Sufferers ranked nausea, pain impact, energy, alertness and constipation; and physicians ranked pain response, CNS effects, nausea, dosage form and constipation in descending order of importance. Sufferers were unwilling to incur severe side-effects to relieve pain, opting out in approximately half of the choice tasks, while physicians were always willing to trade between profiles. The models predicted physicians’ choices well but those of pain sufferers less so. No age, sex or country effects were seen but stronger preferences were expressed by the minority (15%) of physicians treating non-cancer pain, and by the 55% of sufferers who had ever discontinued chronic pain medication and the 41% who reported extreme pain and discomfort in EQ-5D profile responses. Sufferers’ mean pain scores on an 11-point Likert scale were 4.0, 5.7 and 8.6 on their best, average and worst days, respectively. CONCLUSIONS: Online DCEs provide a practical means to compare preferences between physicians and patients, whose concerns overlap substantially but are expressed differently. Sufferers balance pain relief and side-effects. Their refusal to trade between undesirable profiles mirrors high reported rates of treatment discontinuation.