Pharmacokinetic (PK) and pharmacodynamic (PD) data were available from a study of a nasal delivery system for the opioid analgesic fentanyl, together with data on the kinetics of fentanyl in arterial blood in man, and in the lung and brain of sheep. Our aim was to reconcile these data using a physiologically-based population recirculatory PK-PD model, with emphasis on achieving a meta-model that could simultaneously account for the arterial and venous (arm) concentrations of fentanyl, could relate PD effects (pain scores) to the CNS concentrations of fentanyl, and could account for the effect of body size and age on fentanyl kinetics. Data on the concentration gradients of fentanyl across brain, lung and muscle were used to develop sub-models of fentanyl kinetics in these organs. The sub-models were incorporated into a "whole body" recirculatory model by adding additional sub-models for a venous mixing compartment, the liver and gut, the kidney and the "rest of the body" with blood flows and organ volumes based on values for a Standard Man. Inter-individual variability was achieved by allometric scaling of organ size and blood flows, evidence-based assumptions about the effect of weight and age on cardiac output, and inter-individual variability in the free fraction in plasma and hepatic extraction of fentanyl. Post-operative pain scores were found to be temporally related to the predicted brain concentrations of fentanyl. We conclude that a physiologically-based meta-modelling approach was able to describe clinical PK-PD studies of fentanyl whilst providing a mechanistic interpretation of key aspects of its disposition.