Fifteen patients undergoing thoracotomy were given 0.25 or 0.50 mg morphine intrathecally (L2-L3 or L3-L4) for an analgetic and pharmacokinetic study. Administration of morphine at the end of the operation resulted in a highly variable duration of analgesia ranging from 1-20.5 and 1-40 h for the 0.25 and 0.50 mg groups, respectively. Calculation of cumulative consumption pattern of additional analgesics given im indicated a dose-related analgesia lasting around 12 h. Morphine concentrations in the CSF were high and dose dependent. Thus, at 1 h, CSF concentrations (means +/- SEM) were 4,228 +/- 361 ng/ml and 10,447 +/- 1,538 ng/ml for the 0.25 and 0.50-mg groups, respectively. The plasma concentrations generally were very low, i.e., under 1 ng/ml. For the 0.50 and 0.25 mg groups, the terminal elimination half-life in CSF was 175 +/- 9 min and 196 +/- 13 min, respectively: the volume of CSF distribution was 0.88 +/- 0.16 ml X kg-1 and 1.06 +/- 0.17 ml X kg-1, respectively: and the clearance from CSF was 2.81 +/- 0.41 microliter X kg-1 X min-1 and 3.41 +/- 0.55 microliter X kg-1 X min-1, respectively (means +/- SEM). The study indicates that the significant pharmacokinetic parameter related to the long duration of analgesia after intrathecal morphine administration probably is the high CSF concentrations found, since the rate of elimination from CSF is similar to what is reported for morphine in plasma. Furthermore, modulation of nociceptive input in the thoracic region also may be achieved by lumbar administration, but a slower onset should be anticipated.