A better understanding of the natural history of intracerebral haemorrhages (ICH) with cohorts representing the whole spectrum of the disease is necessary to improve treatment. Our aim was to identify potential differences in baseline characteristics and short-term outcomes of patients with non-traumatic ICH, included in a hospital- and in a population-based stroke registry. We compared 373 patients recruited in a university hospital and the last 373 ICH patients included in a population-based registry. Both cohorts included consecutive patients with non-traumatic parenchymal haemorrhages. In the hospital cohort, we collected data from all patients admitted in the emergency room, irrespective of the clinical severity and of the specialist in charge of the patient.In the hospital cohort, patients were younger and more often alcoholic, but these differences may be explained by the younger age and a higher prevalence of alcoholism in this area. Patients also had more frequently hypercholesterolemia, and were more often under antiplatelet therapy. Both cohorts did not differ for intra-hospital casefatality rate.The characteristics of patients included in the hospital cohort were very close to those of patients from a population-based registry, and the differences observed are likely to be explained by differences in the characteristics of the populations in the two areas and different periods of recruitment. Recruiting patients in emergency rooms, and not in stroke units, neurological, or neurosurgical departments, has enabled us to build a cohort of ICH patients representative of the whole spectrum of the disease, with minimised recruitment bias and maximised precision of the variables collected. This cohort may, therefore, provide reliable information on the natural history of ICH.