Recent studies have suggested that morphine-3-glucuronide (M3G) may antagonize the analgesic effects of morphine and morphine-6-glucuronide (M6G). To investigate this hypothesis, steady-state concentrations of morphine, M6G and M3G in serum and cerebrospinal fluid (CSF) were measured in 11 patients receiving chronic morphine therapy (9 orally and 2 subcutaneously) for treatment of cancer-related pain. All patients appeared to have morphine-resistant pain and had elected to proceed to intrathecal bupivacaine or percutaneous cordotomy. Morphine, M6G and M3G concentrations were measured by high-performance liquid chromatography. The concentrations (median and range) for morphine, M6G and M3G in serum were 193 (14–1086) nmol/l, 847 (210–4113) nmol/l and 4553 (1324–24035) nmol/l, respectively, while in CSF concentrations of morphine, M6G and M3G were 200 (21–1461) nmol/l, 115 (30–427) nmol/l and 719 (249–3252) nmol/l, respectively. Median molar ratios of M6G/morphine and M3G/morphine in serum were 3.79 and 22.1, respectively, while in CSF the same ratios were 0.42 and 2.39, respectively. Median molar ratios of M3G/M6G in serum and CSF were 5.84 and 6.61, respectively. The median molar ratios for CSF/serum distribution of morphine, M6G and M3G were 1.23, 0.12 and 0.14, respectively. Thus, despite their relatively poor ability to penetrate into the CSF, the high serum concentrations of M6G and M3G resulted in substantial concentrations of these metabolites in the CSF. Nevertheless, M3G/M6G ratios in our morphine-resistant patients were similar to published values in patients with well-controlled pain, suggesting that the hypothesis that M3G plays a major role in morphine-resistance is not correct.