OBJECTIVES: A study was conducted in new patients registered with tuberculosis (TB) in a rural district of Malawi in order to 1) verify the acceptability of voluntary counselling and testing for human immunodeficiency virus (HIV) infection; 2) describe sexual behaviour and condom use; and 3) identify socio-demographic and behavioural risk factors associated with 'no condom use'. DESIGN: Cross-sectional study. METHODS: Consecutive patients diagnosed with TB between January and December 2000 were offered voluntary counselling and HIV testing (VCT) and were subsequently interviewed. RESULTS: There were 1,049 new TB patients enrolled in the study. Of these, 1,007 (96%) were pre-test counselled, 955 (91%) underwent HIV testing and 912 (87%) were post-test counselled; 43 (4%) patients refused HIV testing. The overall HIV infection rate was 77%. Of all HIV-positive TB patients, 691 (94%) were put on cotrimoxazole. There were 479 (49%) TB patients who reported sexual encounters, of whom only 6% always used condoms. Unprotected sex was associated with having TB symptoms for over 1 month, having had less than 8 years of school education, being single, divorced or widowed or having sex with the same partner. CONCLUSIONS: Offering VCT to TB patients in this setting has a high acceptance rate and provides an opportunity to strengthen and integrate TB and HIV programmes.