We performed a combined analysis of the results from four controlled single-dose relative-potency studies to assess the impact of inferred pain mechanism on the response to an opioid drug. A total of 168 patients received 474 administrations of either morphine or heroin, and we assessed the analgesic response during a 6-hour period with visual analog scales. We summarized this as a total pain relief (TOTPAR) score. Two experienced pain clinicians reviewed information about pain characteristics and designated each case according to the inferred pain mechanism (neuropathic, nociceptive, or mixed) and the degree of confidence in the inferred mechanism (definite versus probable/possible). They grouped the cases as follows: nociceptive pain only (n = 205), neuropathic pain only (n = 49), and mixed (n = 220). We compared pain relief achieved by patients with different mechanisms, with TOTPAR adjusted for significant covariates (duration of prior opioid administration, doses of opioid administered in the previous 48 hours, pain intensity at the start of the study, BUN:creatinine ratio, and dose of administered opioid). The adjusted mean TOTPAR score of the group with any neuropathic pain was significantly lower than that of the group with nociceptive pain only (26.1 versus 20.4, p = 0.02). The score of the group with definite nociceptive pain alone (adjusted mean TOTPAR = 28.0) was significantly higher than scores of the groups with possible/probable nociceptive pain (TOTPAR = 19.9), mixed mechanisms (TOTPAR = 20.2), definite neuropathic pain alone (TOTPAR = 20.6), and possible/probable neuropathic pain alone (TOTPAR = 22.9).(ABSTRACT TRUNCATED AT 250 WORDS)