Setting:: Out-patient tuberculosis clinics in Ilala and Kinondoni Districts, Dar es Salaam, Tanzania. Objective:: To measure the impact on patient adherence to directly observed ambulatory tuberculosis treatment substituting an all-oral treatment regimen for a regimen containing streptomycin. Methods:: The expected and observed attendance of patients during the intensive phase of anti-tuberculosis treatment was measured daily at two out-patient clinics in Dar es Salaam. During the observation period, treatment was changed from a regimen containing streptomycin to an all-oral regimen, and attendance proportions were compared for the three periods during which patients always, sometimes or never received streptomycin during the intensive phase of treatment. Results:: In Kinondoni, an average of 98 patients was expected every day, in Ilala 127. No significant difference was observed in attendance in Kinondoni between periods when patients always (median attendance 95.9%) and never (median 95.7%) received streptomycin injections as part of their intensive phase treatment for tuberculosis. In Ilala, no difference was noted in attendance between the period in which patients received streptomycin for at least part of their treatment (median 91.3%) and the period when ethambutol had fully replaced streptomycin (median 91.8%). Conclusions:: In these two districts of Dar es Salaam, patient adherence to a completely oral treatment regimen was indistinguishable from that to a streptomycin-containing regimen. Given the potential of iatrogenic transmission of HIV and the advantages in reduced staff time and drug costs, the results clearly justify the replacement of streptomycin with ethambutol in Tanzania for new patients receiving an ambulatory rifampicin-containing regimen.