Aims To investigate the pharmacokinetics of morphine, morphine-6-glucuronide (M6G) and morphine-3-glucuronide (M3G) in healthy volunteers after the administration of morphine by subcutaneous bolus injection (s.c.b.) and subcutaneous infusion (s.c.i.) over 4 h, and to compare the results with the intravenous bolus (i.v.) administration of morphine. Methods Six healthy volunteers each received 5 mg morphine sulphate by i.v., s.c.b. and short s.c.i. over 4 h, on three separate occasions, in random order, each separated by at least 1 week. Plasma samples were assayed for morphine, M6G and M3G. Results After i.v. morphine, the concentrations of morphine, M6G and M3G and their pharmacokinetic parameters were similar to those we have observed previously, in other healthy volunteers (when standardized to nmol l x 1 , for a 10 mg dose to a 70 kg subject). After s.c.b. morphine, similar results were obtained except that the median t max values for morphine and M3G were signi®cantly longer than after i.v. morphine (P<0.001 and P<0.05, respectively), with a trend to a longer t max for M6G (P =0.09). The appearance half-lives after s.c.b. morphine for M6G and M3G were also signi®cantly longer than after i.v. morphine (P =0.03 and P<0.05, respectively). Comparison of log-transformed AUC values indicated that i.v. and s.c.b. administration of morphine were bioequivalent with respect to morphine, M6G and M3G. In comparison with i.v. morphine, morphine by s.c.i. was associated with signi®cantly longer median t max values for morphine (P<0.001), M6G (P<0.001) and M3G (P<0.05), and the mean standardized C max values signi®cantly lower than after both i.v. and s.c.b. morphine (morphine P<0.001, M6G P<0.001 and M3G P<0.01 for each comparison). Comparison of log-transformed AUC values after i.v. and s.c.i. morphine indicated that the two routes were not bioequivalent for morphine (logtransformed AUC ratio 0.78, 90% CI 0.66±0.93), M6G (0.72, 90% CI 0.63±0.82), or M3G (0.65, 90% CI 0.54±0.78). A small stability study indicated no evidence of adsorptive losses from morphine infused over 4 h using the infusion devices from the study. Conclusions Although bioequivalence was demonstrated between the s.c.b. and i.v. routes of morphine administration, the bioavailabilities of morphine, M6G and M3G after s.c.i. were signi®cantly lower than after i.v. administration. However, despite this, the study demonstrates that the subcutaneous route is an effective method for the parenteral administration of morphine.