This paper discusses the difficulties in monitoring tuberculosis in an inner-city district health authority. Tuberculosis incidence was reviewed between 1981 and 1989 using several data sources: statutory notifications, hospital activity analysis, death certificates and chest clinic records. The overall notification rate declined but remained higher than national rates; the age-sex distribution was similar to national studies. However, interpretation was limited by the problems of small numbers, the difficulty in obtaining reliable numerator and denominator estimates of ethnic groups, the limited data available about sub-groups of concern such as the homeless, and finally by the incompleteness of the data. Improved methods of data linkage are required to facilitate more complete ascertainment and validation of the diagnosis.