The Social Medicine has faced several difficulties in elaborating methodologies that allow a correlation between the theoretical mark designed and the sensitive reality. Within the investigations in collective health, the nosologic data represent a question of great importance since they constitute the basic information upon which the epidemiologic studies, planning and the organization of health services are based. This study aims at analysing: (a) the spectrum and limits of the nosologic data quality, as registered in medical records at outpatient clinics; (b) the classificatory systems used in the organization of such data. 340 medical records were randomly selected from outpatient clinics at the University Hospital - UFMG from July 1, 87 thru June 30, 88. The registration of the diagnostic hypothesis (DH) and its readibility and presentation were analysed. The DH's were grouped according to the ICD (ICD-9). Those DH's not classifiable by the ICD were also analysed. The study concluded: (a) the nosologic data registered follow the ICD logic; i.e., standard characteristics which follow the clinical reasoning, reflecting the hegemonic conception of the scientific medicine. Despite the continuing revisions, searching for greater specificity, the ICD has incorporated mechanisms which reflect its limitations: the existence of the category Ill Defined Signs, Symptoms, and Morbid States and the presence of a digit in every category that expresses the non-specificity. In the present study, 83% of the records had DH's coded by ICD; 6% of the DH's were in the category Ill Defined Signs, Symptoms, and Morbid States, and 38% did not attain the specificity intended by the ICD; (b) the analysis of those DH's not classifiable, showed that the ICD has not incorporated the characterization of the normal parameters of pediatrics; the analytical nosological conclusions present in psychiatry and the social dimension of disease; (c) none of the alternative classificatory systems analysed answers the limits of ICD.
