Simultaneous pharmacokinetic-pharmacodynamic (PK-PD) models of meperidine in goats were established by utilizing the P3 wave of the cerebral evoked potentials as an analgesic measurement. An effect compartment linked to the central compartment was postulated in the models. The hypothetical drug amount in the effect compartment was related to the observed analgesia through the Hill equation. After intramuscular (i.m., n = 16) and intravenous (i.v., n = 13) dosing (5 mg/kg), the elimination rate constants of meperidine in the effect compartment (Ke0) were 0.3744 +/- 0.2546 and 0.1123 +/- 0.0428 min-1, drug concentrations in the effect compartment generating half maximal analgesia (EC(50)) were 0.70 +/- 0.33 and 0.41 +/- 0.26 microgram/ml, the maximal effects (Emax) were 89.63 +/- 15.63 and 85.92 +/- 9.64%, and the Hill coefficients (S) were 2.61 +/- 1.21 and 2.37 +/- 1.15, respectively. Ke0 and EC(50) with i.m. dosing were significantly greater than with i.v. injection. However, administration route had no influence on S, Emax and the total amount of effect (AUE). The predicted peak effect (Emax) of 64.44 +/- 14.64 and 66.02 +/- 11.51% were achieved at 14.7 +/- 7.4 and 8.5 +/- 2.2 min after i.m. and i.v. dosing, respectively. Peak analgesia appeared much later than peak plasma concentration, but simultaneously with peak CSF level both after i.m. and i.v. dosing. An obvious hysteresis was demonstrated between plasma concentration and analgesic effect. This study demonstrates that meperidine analgesia can be predicted using a PK-PD model, but not by PK data alone. Both i.m. and i.v. administration routes were evaluated kinetically and dynamically.