The diagnosis of pulmonary tuberculosis (PTB) in young children is particularly complex in resource-poor regions where HIV infection is common. This study examines the impact of HIV infection on diagnosis in children with suspected PTB attending Queen Elizabeth Central Hospital, Blantyre. A total of 110 children (4 months-14 years) were studied over a 4-month period. Clinical data were recorded and investigations included Mantoux test, chest X-ray, HIV status (HIV-PCR when younger than 18 months) and sputum, if available. Laryngeal swabs were compared with sputa or gastric aspirates in a subgroup of 60 children. All children were commenced on anti-TB therapy and followed for treatment response. Aware of the clinical overlap between HIV and TB infection, we used more limited criteria than recommended to allocate a final diagnosis following review of all data except HIV status. Final diagnosis included confirmed PTB (n = 8), probable PTB (n = 41), lymphocytic interstitial pneumonitis (n = 10), pulmonary Kaposi sarcoma (n = 3) and bronchiectasis (n = 5). Culture rates of M. tuberculosis were: five (27.8%) of 18 sputa, three (7.1%) of 42 gastric aspirates and four (6.6%) of 60 laryngeal swabs. The HIV infection rate was 70.6% overall and 57.8% in 45 children with confirmed or probable PTB. Although a positive contact history was more common in HIV-infected children, a final diagnosis of confirmed or probable PTB was less common than in HIV-uninfected children (36% vs 63%; p = 0.02). The Mantoux test was positive in 14 (19%) of 72 HIV-infected compared with 15 (50%) of 30 HIV-uninfected children (p < 0.01). A final diagnosis could not be made in 43 (39%) of the study children with suspected PTB, the majority of whom were HIV-infected. HIV-infected children had a significantly poorer response to TB treatment and higher lost-to-follow-up rates.