Abstract : Background Erythropoietin (Epo) is an ubiquitous cytokine. It has endocrine, paracrin and autocrin functions. It improves antiapoptotic mechanisms on all tissues subjected to hypoxic stress. In many animal models of brain trauma but also in other brain injury models, and some human clinical studies, recombinant Human Epo (Epo-rh) has proven neuro-protective properties. The main goal of this work was to improve and incorporate Epo-rh in the pharmacological arsenal of treatment in brain and spinal cord traumatic injuries in human. Material and methods In a first part, to explore the reasons of failure of inclusion in a Spinal Cord Injury (SCI) study to test the thrombo-embolic tolerance and efficacy of Epo-rh, we studied the epidemiology of SCI using the road crash Rhône registry in the period 1997-2006. Then we compared the epidemiological trends of the SCI incidence, associated trauma, mortality and fatality rates in two periods of 6 years: 1995-2001 and 2003-2008. In a second part, due to the 20-fold higher incidence of traumatic brian injury (TBI) in comparison to SCI, we characterized the effects of a moderate (1.6-1.8 atm) lateral fluid percussion injury (LFPI) in order to understand and characterize the pharmacological, antiinflammatory and neuroprotective mechanisms of action of Epo-rh in such a brain injury. Results : The incidence of SCI was 5/100 000 inhabitants in the Rhône area during the 10 year period. However, 40% of the patients died before or immediatly after arriving in hospital facility. This means that, in the 10 years studied, only 57 patients with SCI without associated lesions could be included in the study. Although incidence of severe road crash trauma decreased by 33%, no change in the SCI incidence was observed in the later 6-year period. SCI victims were significantly more aged, polytraumatized, and more frequently using motorcycle, despite mortality increased, non significantly, by 17%. In the experimental LFPI model, we noted histological and genic modifications without significant changes in behavior and MRI-DTI data obtained in the dorsum thalamus and the corpus callosum. If Epo-rh decreased IL-1! and MIP-2 protein concentrations in the cortex, it had no effect either on the 15- to 50-fold increase of the transcrits of IL-1! , TNF" and IL-6 or on the Epo and EpoR genic responses whatever the brain structures analyzed ipsilaterally to the trauma. Bioavailability of Epo-rh in the traumatic cortex was significant 4 hours after intraperitoneal or intravenous injection using at least 3000 UI/kg, was increased in the trauma side and remained elevated for 24h while it disappeared in the non trauma side. Conclusion : it should be difficult to include SCI patients without associated body lesions in Lyon due to the change of the incidence of the main target population. Experimental LFPI of 1.6-1.8 atm corresponds to a mild brain trauma because neither behavioral changes nor MRI- DTI modifications were observed. Such a LFPI induced i) neuronal loss and astroglial reaction in the ipsilateral side, ii) an increase in the inflammatory protein concentrations and associated transcrits, and iii) an increase in Epo and EpoR transcrits. Epo-rh crossed easier the Blood Brain Barrier in the itraumatized side, decreased the IL-1! and MIP-2 protein concentrations in the cortex and had no effect on associated mRNA levels and on Epo system reactivity.