This study was performed to determine whether variations in analgesic responses to intrathecal morphine could be explained by cerebrospinal fluid (CSF) concentrations of morphine metabolites. Twenty-four CSF samples were collected at the beginning, middle and end of treatment periods in seven cancer patients with pain of malignant origin. CSF concentrations of morphine-3,beta-glucuronide (M3G) and morphine-6,beta-glucuronide (M6G) metabolites were measured by gas chromatography/mass spectrometry. Analgesic responses to morphine were estimated concurrent with CSF collection using a visual analog scale representing percentages of pain relief. Effective analgesia was defined as > or = 75% pain relief. CSF concentration of M3G and M6G in the 24 samples were 722 +/- 116 ng/ml and 699 +/- 158 ng/ml, respectively. CSF samples were categorized into two groups: (1) those collected during effective analgesia (N=14), and (2) those collected during ineffective analgesia (N=10). M6G levels detected in group 1 samples (effective analgesia) were significantly greater than those found in group 2 samples (ineffective analgesia) (978 +/- 243 ng/ml vs 309 +/- 68 ng/ml, P<0.05). Intergroup differences in CSF M3G concentrations and M3G/M6G ratios were not significant. It is concluded that CSF M6G may be indicative of effectiveness of analgesia in cancer patients subjected to intrathecal morphine.