This study used positron emission tomography (PET) and [11C]diprenorphine to compare the in vivo distribution abnormalities of brain opioid receptors (OR) in patients with peripheral (n=7) and central post-stroke pain (CPSP, n=8), matched for intensity and duration. Compared with age- and sex-matched controls, peripheral neuropathic pain (NP) patients showed bilateral and symmetrical OR binding decrease, while in CPSP binding decrease predominated in the hemisphere contralateral to pain. In CPSP patients, interhemispheric comparison demonstrated a significant decrease in opioid binding in posterior midbrain, medial thalamus and the insular, temporal and prefrontal cortices contralateral to the painful side. Peripheral NP patients did not show any lateralised decrease in opioid binding. Direct comparison between the central and peripheral groups confirmed a significant OR decrease in CPSP, contralateral to pain. While bilateral binding decrease in both NP groups may reflect endogenous opioid release secondary to chronic pain, the more important and lateralised decrease specific to CPSP suggests opioid receptor loss or inactivation in receptor-bearing neurons. Opioid binding decrease was much more extensive than brain anatomical lesions, and was not co-localised with them; metabolic depression (diaschisis) and/or degeneration of OR neurons-bearing secondary to central lesions appears therefore as a likely mechanism. Central and peripheral forms of NP may differ in distribution of brain opioid system changes and this in turn might underlie their different sensitivity to opiates.