We studied human immunodeficiency virus (HIV)-seroprevalence among children with clinically diagnosed tuberculosis (TB) and compared the clinical features and response to short-term anti-TB therapy of children with and without HIV infection in Santo Domingo, Dominican Republic. Children aged 18-59 months with new-onset, clinically diagnosed TB were tested for HIV antibodies, their clinical features were recorded and their response to a standard 6-month regimen of daily isoniazid and rifampicin with daily streptomycin and pyrazinamide for the first 2 months was assessed. To increase the number of HIV-infected children with TB available for study, we also included children previously known to be HIV infected who developed new-onset TB. Eleven (5.8%) of 189 consecutively enrolled children with clinically diagnosed TB were HIV infected. Fifteen other children with previously documented HIV infection and new-onset TB were available for study, yielding 26 HIV-positive and 178 HIV-negative children with TB. Of these 204 children with clinically diagnosed TB, 25 HIV-positive and 156 HIV-negative children were successfully followed for 6 months or until death. The proportion of HIV-positive children who failed treatment was 6 (29%) of 21 as compared with only 5 (3%) of 156 HIV-negative children [relative risk = 8.9; 95% confidence interval (CI) 2.9, 26.6; p = 0.0004]. HIV-infected children with clinically diagnosed TB are substantially more likely to fail standard treatment for TB than are HIV-uninfected children. If standard treatment regimens are used in such children, response to treatment must be monitored very closely and appropriate changes in the regimen must be made expeditiously.