This study investigates the circumstances surrounding all cases of tub
erculosis found at autopsy in Auckland during the period 1975-1992. Ca
ses were placed in one of six categories according to the medical hist
ory and autopsy findings. Of the 34 cases in the study 22 were clinica
lly undiagnosed prior to autopsy. Of these, six exhibited respiratory
symptoms that should have led to a diagnosis and three reported unexpl
ained chronic symptoms such as weight loss for which tuberculosis shou
ld at least have been considered. Only 10 cases were correctly diagnos
ed before death. The autopsy plays an important role in finding undiag
nosed cases of tuberculosis, as patient contacts may then be followed
up. The vital role of post-mortem examination in medical audit and pub
lic health is, once again, demonstrated.