Introduction: Globally, one in four persons infected with the human immunodeficiency virus (HIV) who are living with the acquired immunodeficiency syndrome (AIDS) will die of tuberculosis (TB). The estimated number of HIV-infected children who die of tuberculosis, especially extrapulmonary TB (EPTB), in Haiti, is only loosely based on facts or investigation. This study proposes to describe demographics of children with EPTB in a pediatric TB hospital in Haiti. The objectives are two-fold. The first objective is to describe the population of children discharged from Grace Children's Hospital with a confirmed diagnosis of tuberculosis overall, and broken down by whether or not the child had an extrapulmonary manifestation of the TB disease. Specifically, we describe the demographic characteristics and the prevalence of HIV and other co-morbidities of the children, in-hospital mortality, and the diagnostic tools used to determine TB infection including the sputum test, and th e documentation of family members also infected. As part of the descriptive process, by examining those with only pulmonary TB (PTB) and EPTB separately, we investigate if they appear to be different sub-populations based on demographic characteristics and clinical measures. The second objective of this work is to determine if there is a positive association between HIV infection and the EPTB manifestation in children with a confirmed diagnosis of TB, both crudely and after adjusting for demographic variables and co-morbidities. Methods: A cross-sectional study design was used to review medical charts of clinically diagnosed pediatric TB cases for a five-year period (January 1, 1999 -- December 31, 2003). This included 492 pulmonary TB and 210 extrapulmonary TB cases. Variables measured included clinical measures and demographic characteristics. Results: Data for 615 hospitalized, clinically diagnosed pediatric TB cases were reviewed. There were 315 (51.4%) males and 298 (48.6%) females with a mean age of 5.40 years (range 0.17 - 14 years), with 214 (37.9%) of the patients aged 0-2 years. Percent males were 47.8% and 57.9% in PTB and EPTB groups respectively (p<0.05). One hundred and seven (17.4%) of patients were HIV positive. Three hundred eighty-eight (63%) of the patients had one or more additional co-morbidities: [anemia 299 (48.6%), intestinal parasites 93 (15.1%), malaria 58 (9.4%) and gastroenteritis 19 (3.1%)]. Nearly 85% of the children were undernourished. Eighty-three child patients (13.5%) died in the hospital. Children with EPTB were much more likely to be over the age of two (74% vs. 56% in PTB group), resulting in a highly significant Chi-square stati stic. The overall difference in mean age, however, was only borderline significant with children with EPTB being slightly older [p=0.059] and age was only weakly associated with TB group. They were much less likely to be HIV positive (8.6% vs. 22%, p<0.01). Children in the EPTB group were somewhat less likely to die in the hospital (10.0% vs. 15.4%, p=0.066). The OR was greater than 4 for HIV and was greater than 2 for poor nutrition status [p<0.01 for each]. Conclusion: There was no association in this model between EPTB and mortality. The apparent univariate association between EPTB and reduced mortality can be explained by lower prevalence of HIV and poor nutrition status in this sub-sample. This study has implications for hospital-based pediatric TB diagnosis and epidemiology in resource-poor countries.