In January 2002 the Communicable Disease Surveillance Centre (CDSC) introduced outcome reporting for tuberculosis 12 months after start of treatment. To determine whether outcome could change with longer and more detailed follow up, we examined this at 12 months and two years for a cohort notified in 2000 at a central London hospital. At 12 months 61/94 (65%) notified patients had completed treatment, 10 (16%) had died, 11 (12%) were lost to follow up, six (6%) were transferred to another service in the UK, four (4%) were still on treatment, and treatment for one (1%) patient had been stopped. After a mean follow up of two years, 65 (69%) had completed treatment. In this London service with a high prevalence of HIV infection (at least 23%), low treatment success was due to deaths, transfers and losses to follow up. The last of these was often due to patients returning to their country of origin (7 of 11). Tuberculosis was the primary or contributing cause of death in at least 4/94 (4%) cases. Completion rates need to be interpreted with caution particularly in specialist units with highly mobile populations. This has implications for national targets as well as for models of care.